Fifty-nine consecutive patients admitted for colonoscopy were randomized to receive polyethylene glycol or sodium picosulphate. Patients expressed their opinion in a questionnaire and the endoscopists, blinded to the preparation, assessed the cleanliness of different segments of the colon. There was no statistically significant difference in the taste-acceptability of the preparations, frequency of nausea, abdominal pain, peri-anal soreness or sleep disturbance between the two groups. Polyethylene glycol caused vomiting in 13% of patients while this was absent in those who received sodium picosulphate (P less than 0.05). The average number of stools passed was 12.4 in the polyethylene glycol and 8.6 in the sodium picosulphate groups; mean difference 3.8 (95% C.I. 0.7-6.9) with P less than 0.02. The overall cleanliness of the colon was better in the polyethylene glycol group (P = 0.002) as judged by the blinded colonoscopist. There was less delay (P = 0.06) and more completed colonoscopies (P = 0.01) in this group. Polyethylene glycol was a better preparation in all segments of the colon except the rectum. We conclude that polyethylene glycol is the choice of the colonoscopist and should be given to all patients; sodium picosulphate would be a good alternative if patients are intolerant. If a limited colonoscopy or flexible sigmoidoscopy is intended, sodium picosulphate may be preferred because of its acceptable efficacy and slightly advantageous side-effect profile.