AIMS: The purpose of the present study was to compare the effectiveness of instruction to contract the pelvic floor muscles (PFM), the transversus abdominis (TrA), and the TrA + PFM visualized as displacement of the pelvic floor by ultrasound. MATERIALS AND METHODS: Twenty female physical therapists, mean age 41.1 years (range 26-56) participated in the study. A 3.5 MHz 35 mm curved linear array ultrasound transducer (Dornier Medtech) was placed in the mid-sagittal plane immediately suprapubically, angled at 15-30 degrees from the vertical depending on subcutaneous fat and anatomical variations, to image the pelvic floor. Six trials of three maneuvers in random order were performed: contraction of PFM, TrA, and TrA + PFM. RESULTS: In spite of correct contractions assessed by palpation and clinical observation, one subject demonstrated a downward movement of the pelvic floor during PFM contraction on ultrasound. Six subjects (30 ) showed a downward movement during a TrA- contraction, and two during the combined TrA + PFM contraction. Instruction to contract PFM produced significantly greater mean displacement: 11.2 mm (95 CI 7.2-15.3) than TrA 4.3 mm (95 CI -0.2-8.8), P <0.01, and combination: 8.5 mm (95 CI 5.2-12), P = 0.04. Hence, instruction of PFM contraction produced a 61.6 greater displacement of the pelvic floor in the correct direction than a TrA contraction. CONCLUSIONS: It is concluded that ultrasound is a more valid method than palpation and clinical observation to assess PFM function, and that instruction to contract the PFM produces a significantly more effective pelvic floor muscle contraction than instruction to perform a TrA contraction.