In this study, we tested two hypotheses. First, that consultation-liaison (C-L)-trained psychiatry residents would order constant observation (CO) less frequently than psychiatry residents untrained in C-L. Second, we predicted that CO would be ordered less frequently under circumstances when experienced C-L psychiatry attending and fellows would be available to supervise psychiatry residents training in C-L. We reviewed a total of 138 consultations during a 6-month period. Constant observation was recommended in 31 cases (22.5%). Consultations were done by residents who had received training in C-L psychiatry (n=34) and by residents who were not trained in C-L (n=34). Residents not trained in C-L had a significantly higher percentage of CO orders (44.1%) compared to those trained in C-L (15.4%) (chi(2)=12.1, df=1, P<0.001). Because C-L-trained residents provided regular-hour and after-hour consults while residents without C-L training provided only after-hour consults, we also separately analyzed data from the 102 after-hour cases. We again found that residents with C-L training had a significantly lower rate of ordering CO (22.1%) than those who had not yet received C-L psychiatry training (44.1%) (chi(2)=5.31, df=1, P<0.05). We also found that C-L-trained residents ordered CO less frequently during regular hour consults (2.8%) when experienced staff are available in supervision compared to after hours (22.1%) (chi(2)=6.72, df=1, P<0.01). Our findings suggest that training in C-L psychiatry has a significant impact on the use of constant observation for patients in the general hospital thereby reducing the cost of care.