The objective was to compare the summary statistics for fatal antidepressant and antipsychotic drug concentrations mined by different approaches from post-mortem (PM) toxicological data. Reference concentrations of drugs play a major part in the interpretation of PM toxicology results. However, PM drug concentrations are not necessarily the same as those at the time of death, and consequently standard tables of clinical plasma concentration data are of limited value. Compilations of PM concentrations collected from literature sources are often heterogeneous in terms of the sampling site, the analytical methods used, and the number of cases involved. However, there are also published summary statistics for drug concentrations in PM femoral venous blood, based on extensive databases from centralized laboratories with well-established and high-quality working practices. The present study establishes concentration distributions for a number of antidepressant and antipsychotic drugs, specified as a principal finding in fatal poisoning (principal finding approach). These levels are then compared with those obtained using two previous approaches: according to the all-causes-of-death (Ketola RA, Drug Test Anal, 2019, 11, 1326–1337) or according to the grouped causes of death, aka “the Druid approach”, where the reference information is subdivided into poisonings by one specific substance only, multi-substance poisonings and non-poisoning controls (Reis M, J Anal Toxicol, 2007, 31, 254-264; Söderberg C, Forensic Sci Int, 2016, 266, 91-101). Summary statistics for fatal antidepressant and antipsychotic drug concentrations in PM femoral venous blood were generated according to the principal drug finding. In most cases, also other drugs were implicated in the cause of death. The data were mined between 2000–2020 from the national PM toxicology database, maintained by the Finnish Institute for Health and Welfare, in which all results of the toxicological analyses in medico-legal investigations and information from the respective death certificates are collected. From this material, those antidepressant and antipsychotic drugs that had been quantitatively analysed at least 10 times were selected. Concentration percentiles (10th, 50th, 90th, 95th, 97.5th) were calculated for 16 different antidepressant ( N = 1999) and for 7 different antipsychotic drugs ( N = 1134), defined as principal findings of fatal poisoning by forensic pathologists. For instance, for the antidepressant amitriptyline the respective concentrations were 0.66, 2.0, 8.3, 13 and 25 mg/L, and for the antipsychotic levomepromazine (methotrimeprazine) 0.40, 1.5, 5.3, 10 and 16 mg/L. A comparison with other data mining approaches will be presented in visual form. A clear correspondence was found between, for example, the 90th percentile concentration level of the present principal finding approach and the 97.5th percentile concentration level of the all-causes-of-death approach. Similarly, a correspondence was found between, for example, the median concentration level of the present principal finding approach and the median concentration level of the multi-substance poisoning group in the Druid approach. The novel data presented in this study followed by a systematic comparison with data from previous studies helps rationalize the use of reference concentrations from various sources in the interpretation of potential poisoning deaths. Further data mining for additional drugs will be markedly easier in the principal finding approach than in the Druid approach, provided that the principal findings are specified in advance in the database as is in our case. Despite the proven usefulness of reference concentrations, interpretation must always be made on a case-by-case basis, taking into account e.g. background information, autopsy findings, and substance interactions. A forensic toxicologist must have the evidence-based PM concentration information at his or her disposal when assisting a forensic pathologist in determining the cause of death or being called to testify in court.