Accountability has become a concern for public mental health programs, including the Hawaii State Mental Health Division which relies on the Multi-State Information System (MSIS) for program planning and evaluation. Accuracy in reporting and agreement in clinical judgment on the MSIS has been assumed but never tested until an analogue study was undertaken to test these assumptions. A fictitious intake interview was presented to 115 clinicians throughout the State Mental Health Division. Based on the protocol, the clinicians were asked to complete MSIS Admission Form items as well as a Global Assessment Scale rating and an item concerning treatment expectations. Inaccuracies in recording factual data as well as major disagreement in clinical judgments were found, with an error rate of 3 to 18% occurring on factual items such as marital status, education and income. On a 38-item problem checklist, there was substantial variability in the number and type of problems checked. In assessing severity of condition, the distribution of ratings on the 100-point Global Assessment Scale indicated one-third of the ratings above 60 (mild), one-third between 51 and 60 (moderate), and the remaining third in the lower half of the scale (severe). There was little agreement in anticipated treatment duration ranging from 4 to over 50 visits. Results suggest that inaccuracies in such data may have serious adverse consequences for both the clients and the mental health services. (Author)