Transversus abdominis plane block provides postoperative analgesia following abdominal surgery by targeting thoracolumbar nerves between the internal oblique and transversus abdominis muscles. Posterior and subcostal approaches using ultrasound guidance have been described. However there have been inconsistent results in relation to the extent of the sensory block. This observational study evaluated the distributions of sensory block following either a posterior or subcostal approach and the quality of analgesia achieved. Following ethics committee approval, 50 patients undergoing minimally invasive and major abdominal surgery were recruited. A total of 81 transversus abdominis plane blocks were performed preoperatively under real-time ultrasound guidance. Postoperatively, patients received multimodal analgesia including morphine via patient-controlled pumps. Ninety-eight percent of patients had some degree of demonstrable sensory block and the dermatomal spread differed between posterior and subcostal approaches (P < 0.001). The posterior approach produced a median sensory block of three dermatomal segments (interquartile range 2 to 4), the most cephalad being T10 (interquartile range T9 to T10), while the subcostal approach blocked a median of four segments (interquartile range 3 to 5), the most cephalad being T8 (interquartile range T7 to T9, P < 0.001). Maximum dermatomal block distribution was observed at 30 minutes and usually regressed by 24 hours. Median cumulative morphine consumption was 40.8 mg (interquartile range 17 to 50 mg) at 24 hours. Median pain scores at rest and with coughing were 20 (interquartile range 10 to 35) and 50 (interquartile range 29 to 67) respectively at 24 hours. The posterior approach appears to be more appropriate for lower abdominal surgery and the subcostal approach better suited to upper abdominal surgery. Whichever approach is used, transversus abdominis plane block is only one component of a multimodal analgesic technique.