88 men undergoing elective prostatectomy were investigated by transrectal and transabdominal ultrasound. A repeat examination was performed by a second observer in 28 patients. The aims were to determine the correlation between transrectal and transabdominal ultrasonic estimation of prostatic volume and the degree of observer error. Transabdominal estimation of prostatic volume correlated well with the transrectal method (p less than 0.001) and good agreement between the two observers was found with both methods (p less than 0.001). In the individual patient, however, a wide variation in prostatic volume was found between the two observers and the two methods.