BACKGROUND: We performed a prospective, randomized study to investigate the relations between BIS (bispectral index, BISTM monitor), AAI (mid-latency AEP, A-line ARX index) and depth of sedation, and assessed two different memory functions at OAA/S score 2 under spinal anesthesia. METHODS: 44 ASA physical status 1-2 patients, scheduled for elective total knee replacement arthroplasty were randomly allocated to either the BIS group (22) or the AEP group (22). Under spinal anesthesia, patients were sedated with propofol TCI to OAA/S score 2. BIS and AAI at each OAA/S score were measured and patients were maintained at an OAA/S score 2. A specific picture was shown during the operation. At PACU, an interview was done to determine whether patients recalled specific picture (picture recall) or any intra-operative noises or scenes (op recall). RESULTS: BIS decreased as OAA/S score reduced from 5 to 2 (P < 0.05), but AAI could not differentiate between an OAA/S score 3 and 2 (P = 0.0690). There were correlations of 0.827 and 0.610, respectively, (Spearman's rho) between BIS and OAA/S scores and between AAI and OAA/S scores. BIS was higher at all OAA/S scores (P < 0.05), and the standard deviations of BIS were smaller. There was a statistically significant difference between the recalls (P = 0.0253). CONCLUSIONS: Both BIS and AAI provided information about the depth of sedation, but BIS was more accurate. The difference between 'op recall' and 'picture recall' requires further investigation.