Background, aim, and scope In response to concerns about chemical substances that can alter the function of endocrine systems and may result in adverse effects on human and ecosystem health, a number of in vitro tests have been developed to identify and assess the endocrine disrupting potential of chemicals and environmental samples. One endpoint that is frequently used in in vitro models for the assessment of chemical effects on the endocrine system is the alteration of aromatase activity (AA). Aromatase is the enzyme responsible for converting androgens to estrogens. Some commonly used aromatase assays, including the human microsomal assay that is a mandatory test in US-EPA's endocrine disruptor screening program (EDSP), detect only direct effects of chemicals on aromatase activity and not indirect effects, including changes in gene expression or transcription factors. This can be a problem for chemical screening initiatives such as the EDSP because chemicals can affect aromatase both indirectly and directly. Here we compare direct, indirect, and combined measurements of AA using the H295R cell line after exposure to seven model chemicals. Furthermore, we compare the predictability of the different types of AA measurements for 17β-estradiol (E2) and testosterone (T) production in vitro. Materials and methods H295R cells were exposed to forskolin, atrazine, letrozole, prochloraz, ketoconazole, aminoglutethimide, and prometon for 48 h. Direct, indirect, and combined effects on aromatase activity were measured using a tritiated water-release assay. Direct effects on aromatase activity were assessed by exposing cells only during the conduct of the tritium-release assay. Indirect effects were measured after exposing cells for 48 h to test chemicals, and then measuring AA without further chemical addition. Combined AA was measured by exposing cells prior and during the conduction of the tritium-release assay. Estradiol and testosterone were measured by ELISA. Results and discussion Exposure to the aromatase inhibitors letrozole, prochloraz, ketoconazole, and aminoglutethimide resulted in greater indirect aromatase activity after a 48-h exposure due to presumed compensatory mechanisms involved in aromatase activity regulation. Forskolin and atrazine caused similar changes in hormone production and enzyme profiles, and both chemicals resulted in a dose-dependent increase in E2, T, and indirect AA. Neither of these two chemicals directly affected AA. For most of the chemicals, direct and combined AA and E2 were good predictors of the mechanism of action of the chemical, with regard to AA. Indirect aromatase activity was a less precise predictor of effects at the hormone level because of presumed feedback loops that made it difficult to predict the chemicals' true effects, mostly seen with the aromatase inhibitors. Further, it was found that direct and indirect AA measurements were not reliable predictors of effects on E2 for general inducers and inhibitors, respectively. Conclusions Differential modulation of AA and hormone production was observed in H295R cells after exposure to seven model chemicals, illustrating the importance of measuring multiple endpoints when describing mechanisms of action in vitro. Recommendations and perspectives For future work with the H295R, it is recommended that a combination of direct and indirect aromatase measurements is used because it was best in predicting the effects of a chemical on E2 production and its mechanism of action. Further, it was shown that direct AA measurements, which are a common way to measure AA, must be used with caution in vitro.