The parameters used were blood unconjugated morphine, blood, brain, and liver total morphine, sex, age, frequency of use, time of death after injection, cause of death, and presence of other drugs. The computer programs used were Expert 4, BEAGLE, and Knowledge Maker. The interpretation was defined as estimating the dose, response, and the time after drug dosing. The AI programs were used to advise on time and response outcomes for cases, to calculate the probability of the estimates being true, to develop rules for interpretation of morphine-involved cases, and to diagram a decision tree. The AI programs had a success rate of 70-90 percent to classify the cases as to response and time. The success rate for individual cases was proportional to the program-estimated probability; the three programs all found the case parameters of most value in predicting response to be blood unconjugated morphine, blood total morphine, and liver total morphine. Results indicate that inexpensive AI programs available commercially for personal computers can be useful in interpretation in forensic toxicology. 10 tables, 3 figures, 14 references. (Author abstract modified)