Aim: To identify the effects of local recommendations of pharmacogenetic testing in psychiatry with respect to treatment costs. Methods: Based on Danish patient registers, individual treatment costs within a 365-day period at three psychiatric hospitals recommending and using pharmacogenetic testing is compared retrospectively with treatment costs at other Danish psychiatric hospitals using alternate treatment strategies. Primary outcome of interest is total direct costs analyzed by multilevel modelling. Secondary outcome measures are healthcare consumption within specific sectors analyzed by Tobitregressions. Results: Costs among patients treated at hospitals recommending and using pharmacogenetic testing were not found to be statistically significantly different from costs among patients treated at sites using alternate treatment strategies. In spite of recommendations to test all patients the uptake of the test was, however, low (26—31 %). Treatment practice using routine therapeutic drug monitoring (in Ãrhus) shows a trend towards lower costs. Conclusions: Based on this natural experiment we were not able to document statistically significant differences in total costs between treatment sites that had guidelines recommending pharmacogenetic testing, relative to sites without such guidelines, over a period of one year. However, guidelines of pharmacogenetic testing and possibly also therapeutic drug monitoring seem to lead to reductions in costs for primary care services. In the case of the former, reductions do, however, seem to be outweighed by increases in costs for psychiatric and non-psychiatric inpatient stays. In conclusion, no statistically significant differences in total direct costs across sites with different treatment strategies were found.