Laparoscopic-assisted surgery for colorectal cancer has been widely adopted without data from large-scale randomised trials to support its use. We compared short-term endpoints of conventional versus laparoscopic-assisted surgery in patients with colorectal cancer to predict long-term outcomes. Between July, 1996, and July, 2002, we undertook a multicentre, randomised clinical trial in 794 patients with colorectal cancer from 27 UK centres. Patients were allocated to receive laparoscopic-assisted (n=526) or open surgery (n=268). Primary short-term endpoints were positivity rates of circumferential and longitudinal resection margins, proportion of Dukes' C2 tumours, and in-hospital mortality. Analysis was by intention to treat. This trial has been assigned the International Standard Randomised Controlled Trial Number ISRCTN74883561. Six patients (two [open], four [laparoscopic]) had no surgery, and 23 had missing surgical data (nine, 14). 253 and 484 patients actually received open and laparoscopic-assisted treatment, respectively. 143 (29%) patients underwent conversion from laparoscopic to open surgery. Proportion of Dukes' C2 tumours did not differ between treatments (18 [7%] patients, open vs 34 [6%], laparoscopic; difference -0.3%, 95% CI -3.9 to 3.4%, p=0.89), and neither did in-hospital mortality (13 [5%] vs 21 [4%]; -0.9%, -3.9 to 2.2%, p=0.57). Apart from patients undergoing laparoscopic anterior resection for rectal cancer, rates of positive resection margins were similar between treatment groups. Patients with converted treatment had raised complication rates. Laparoscopic-assisted surgery for cancer of the colon is as effective as open surgery in the short term and is likely to produce similar long-term outcomes. However, impaired short-term outcomes after laparoscopic-assisted anterior resection for cancer of the rectum do not yet justify its routine use.