'The relevant substances in scents of human corpses are not clearly identified and hence, a confirmation of results obtained by cadaver dogs does not exist and at this time, cannot be reproduced by means of an instrumental analysis. Therefore, the designation by a dog that a deceased body has potentially contaminated a surface may be used only as a strong indicative tool during the investigative process of a homicide case. This study demonstrated the fact that only very few searches by trained dogs resulted in a false-positive signal and hence, emphasized the necessity of employing these dogs to enhance the accuracy and reliability of homicide investigations worldwide.

As no other analytical confirmation is possible at this time, a positive signal by a trained cadaver dog should not be used as the sole of evidentiary piece in court. These dogs remain ‘‘only’’ an excellent search tool. Furthermore this investigative
study produced excellent results and as Schoon [17] has already determined, conditions of an experimental design do significantly differ in many aspects from an actual crime scene. Thus, our results may not be extrapolated to all environments
encountered by investigative law enforcement units.

'The olfactory sense of dogs is known to be highly sensitive and allows for a near perfect discrimination of different odors [7]. Cadaver dogs are trained to identify and signal upon encountering scents associated with deceased tissue. A monosubstance scent such as cadaverine or putrescine does not initiate a positive signal by these dogs. Other components of this specific scent must be present so that these dogs give a positive signal. Our investigation was based on using scents that were obtained from individuals with a postmortem interval of less than 3 h—an interval in which some organs and many cells of the human organism are still vital and have not yet started the process of putrefaction. Hence, further research is necessary that may discern different components of scents associated with various postmortem and vitality intervals. The olfactory sense of trained dogs is highly sensitive and the dog’s ability to smell seems to be far superior to the level of science today...................................

.........The most interesting question of all remains: that of how long must an
individual be dead for his/her scent to be detectable by a trained cadaver dog? Answering this pertinent question was not part of our investigation, but we can point out that a postmortem interval of 2 h seems to be a safely recognizable interval for the detection of deceased tissue by trained cadaver dogs. '

SOURCE
Cadaver dogs—A study on detection of contaminated carpet squares
Oesterhelweg, L., et al
Forensic Science International 174 (2008) 35–39
[available online at <!-- m -->http://www.sciencedirect.com<!-- m -->] "