How Cadaver dogs are trained
What the dog is actually smelling
A decomposing human body releases hundreds of volatile organic compounds, and the mix shifts continuously as decomposition progresses. There is no single “death scent” and no single compound that means “human.”
Two compounds are constantly mentioned in popular accounts — cadaverine and putrescine. They are real, they are present, and they are worth knowing about. But they are not specific to humans. Both are diamines produced during the breakdown of animal tissue, which is exactly why a dog trained on those two compounds alone would be a poor human-remains dog. It is also why pseudo-scent products built around them should be treated with caution rather than confidence.
"Some have said that when Humans decompose, they eventually smell like animal remains, and NOTHING could be farther from the truth." -- Michael Hadsell, President and Founder, Peace River K9 Search and Rescue Association.
What the research actually shows is a sparse and variable signature. In one study of VOCs recovered from human bone, 296 compounds were identified across the sampling, and no single compound appeared in even half the samples — the most frequent showed up in about a third. Roughly 40 percent were detected exactly once. A dog trained on a narrow slice of that signature is trained on a narrow slice.
Species specificity is real, though, and it has been measured. Work on bone VOCs found 168 compounds from human bone, of which only eight were shared with pig and 38 with deer — and found that trained dogs can discriminate human bone from non-human bone. That is a good argument for human-sourced training material and against relying on animal proxy.
How the Odor is Taught
Imprinting
The dog is introduced to the target odor and learns to associate it with reward. Early sessions use simple, low-complexity presentations, and difficulty is added deliberately — not all at once — as the dog demonstrates that it has the concept.
Training aids
ANSI/ASB Std 076-2026 sets out six categories of training aid, and it requires that certification testing use the required aids. It does not require a team to own them. That distinction matters in practice: certifying on site at a facility that holds donated remains can satisfy the standard's aid requirement, its rule that certification aids not be aids the team trained on, and its bar on certifying where the team routinely trains — all at once.
076 does not address alternative or synthetic training aids at all. A proposed standard covering their verification is in development and has not been published. Until it is, any team using pseudo-scent should be able to say what it is validated against, and any agency should feel free to ask.
Distractors and blanks
A dog that has only ever searched areas containing the target has not been taught the hardest part of the job. Training must include areas with no target odor at all — blanks — and areas containing competing material such as animal remains, food, and household chemicals. A dog that responds in a blank area has told you something important, and so has a training program that never runs one.
Blind and double-blind work
In a blind exercise, the handler does not know where the aid is. In a double-blind, nobody present knows. This is not a formality. In a controlled study of certified detection teams in which no target odor was present anywhere, handler belief — cued by nothing more than a paper marker — shifted alert locations more than planted decoy scent did. Every alert in that study was wrong, because there was nothing there to find. Double-blind work is the only way to see past that, and it is why 076 changed the rules on odor recognition testing so the handler no longer knows how many targets are in play.
The trained final response
When the dog identifies the source, it performs a trained behavior to signal the handler. Broadly, there are two styles:
Note on a common confusion: “Live find” is not a type of cadaver dog alert. It is a different discipline — a dog trained to locate living people. Some dogs are cross-trained for both and give a different response to each, which is demanding to build and maintain. But an alert style and a search discipline are two different axes, and conflating them is a reliable sign that a description was written by someone outside the field.
Just as important as the final response is the change of behavior that precedes it — the shift in the dog's carriage, pace, breathing, and search pattern as it comes into odor. Reading that is what most of a handler's work actually is. Under 076, a response that cannot be confirmed is recorded as a non-productive response, not as a failure and not as a false alert.
Certification
Certification is a snapshot of a team on one day. It is necessary, and it is not sufficient — the training record behind it is what shows the team is still where the certificate says it was.
Under 076, a certification that means something has three properties an agency can check:
A certificate issued by a team's own trainer, on the team's own aids, at the team's own field, does not meet the standard no matter what it says on its face.
Maintenance and records
Skills decay. Training continues throughout the dog's working life, across varied terrain, weather, source age, source size, and burial depth, because a dog that has only worked one kind of problem will be reliable only on that kind of problem.
076 requires four separate record types — training, certification, deployment, and maintenance — with 70 fields between them. PRSAR keeps them and makes them available to the authority having jurisdiction on request.
The Handler is Half the Team
076 specifies 21 handler training topics. The dog supplies the nose; nearly everything else that determines whether a search produces usable evidence is the handler — search planning, scene preservation, knowing when odor is unavailable rather than absent, accurate records, and honest reporting. Cognitive bias belongs on that list as squarely as scent theory does. A handler who cannot describe how their own expectations might shape their dog's behavior has not finished training.
A decomposing human body releases hundreds of volatile organic compounds, and the mix shifts continuously as decomposition progresses. There is no single “death scent” and no single compound that means “human.”
Two compounds are constantly mentioned in popular accounts — cadaverine and putrescine. They are real, they are present, and they are worth knowing about. But they are not specific to humans. Both are diamines produced during the breakdown of animal tissue, which is exactly why a dog trained on those two compounds alone would be a poor human-remains dog. It is also why pseudo-scent products built around them should be treated with caution rather than confidence.
"Some have said that when Humans decompose, they eventually smell like animal remains, and NOTHING could be farther from the truth." -- Michael Hadsell, President and Founder, Peace River K9 Search and Rescue Association.
What the research actually shows is a sparse and variable signature. In one study of VOCs recovered from human bone, 296 compounds were identified across the sampling, and no single compound appeared in even half the samples — the most frequent showed up in about a third. Roughly 40 percent were detected exactly once. A dog trained on a narrow slice of that signature is trained on a narrow slice.
Species specificity is real, though, and it has been measured. Work on bone VOCs found 168 compounds from human bone, of which only eight were shared with pig and 38 with deer — and found that trained dogs can discriminate human bone from non-human bone. That is a good argument for human-sourced training material and against relying on animal proxy.
How the Odor is Taught
Imprinting
The dog is introduced to the target odor and learns to associate it with reward. Early sessions use simple, low-complexity presentations, and difficulty is added deliberately — not all at once — as the dog demonstrates that it has the concept.
Training aids
ANSI/ASB Std 076-2026 sets out six categories of training aid, and it requires that certification testing use the required aids. It does not require a team to own them. That distinction matters in practice: certifying on site at a facility that holds donated remains can satisfy the standard's aid requirement, its rule that certification aids not be aids the team trained on, and its bar on certifying where the team routinely trains — all at once.
076 does not address alternative or synthetic training aids at all. A proposed standard covering their verification is in development and has not been published. Until it is, any team using pseudo-scent should be able to say what it is validated against, and any agency should feel free to ask.
Distractors and blanks
A dog that has only ever searched areas containing the target has not been taught the hardest part of the job. Training must include areas with no target odor at all — blanks — and areas containing competing material such as animal remains, food, and household chemicals. A dog that responds in a blank area has told you something important, and so has a training program that never runs one.
Blind and double-blind work
In a blind exercise, the handler does not know where the aid is. In a double-blind, nobody present knows. This is not a formality. In a controlled study of certified detection teams in which no target odor was present anywhere, handler belief — cued by nothing more than a paper marker — shifted alert locations more than planted decoy scent did. Every alert in that study was wrong, because there was nothing there to find. Double-blind work is the only way to see past that, and it is why 076 changed the rules on odor recognition testing so the handler no longer knows how many targets are in play.
The trained final response
When the dog identifies the source, it performs a trained behavior to signal the handler. Broadly, there are two styles:
- Passive: the dog sits, downs, or stands and freezes at the source. Standard for human remains work, because it does not disturb the scene.
- Active: the dog barks, scratches, or digs. Useful in rubble and some water work; generally wrong for a site that may be excavated as evidence.
Note on a common confusion: “Live find” is not a type of cadaver dog alert. It is a different discipline — a dog trained to locate living people. Some dogs are cross-trained for both and give a different response to each, which is demanding to build and maintain. But an alert style and a search discipline are two different axes, and conflating them is a reliable sign that a description was written by someone outside the field.
Just as important as the final response is the change of behavior that precedes it — the shift in the dog's carriage, pace, breathing, and search pattern as it comes into odor. Reading that is what most of a handler's work actually is. Under 076, a response that cannot be confirmed is recorded as a non-productive response, not as a failure and not as a false alert.
Certification
Certification is a snapshot of a team on one day. It is necessary, and it is not sufficient — the training record behind it is what shows the team is still where the certificate says it was.
Under 076, a certification that means something has three properties an agency can check:
- The aids used were not aids the team had trained on.
- The location was not where the team routinely trains.
- The certifying official was independent of the team's own training program.
A certificate issued by a team's own trainer, on the team's own aids, at the team's own field, does not meet the standard no matter what it says on its face.
Maintenance and records
Skills decay. Training continues throughout the dog's working life, across varied terrain, weather, source age, source size, and burial depth, because a dog that has only worked one kind of problem will be reliable only on that kind of problem.
076 requires four separate record types — training, certification, deployment, and maintenance — with 70 fields between them. PRSAR keeps them and makes them available to the authority having jurisdiction on request.
The Handler is Half the Team
076 specifies 21 handler training topics. The dog supplies the nose; nearly everything else that determines whether a search produces usable evidence is the handler — search planning, scene preservation, knowing when odor is unavailable rather than absent, accurate records, and honest reporting. Cognitive bias belongs on that list as squarely as scent theory does. A handler who cannot describe how their own expectations might shape their dog's behavior has not finished training.
