Objective: To assess the efficacy and safety of zolpidem administered on an 'as needed' basis for 4 weeks in primary insomnia. Design: Multicentre, double-blind, randomised, placebo-controlled study. Setting: 58 primary care centres in France. Patients and Participants: 245 individuals with primary insomnia. Methods: Following a placebo run-in period, patients were given four tablets containing either placebo or zolpidem 10mg and were instructed to take one tablet on the first two nights, the two remaining tablets to be taken when required throughout the week. This was followed by a 3-week period during which patients were provided with placebo or zolpidem but advised to use as few tablets as possible. Results: Premature withdrawals due to inefficacy were significantly more common with placebo than with zolpidem (5% vs 0%: p = 0.01). At the end of the study, total sleep time was increased in both groups, the difference being slightly more marked in the zolpidem group (74.6 vs 63.2 minutes; p = nonsignificant). Zolpidem was significantly (p < 0.05) more effective than placebo with respect to sleep quality, daytime drowsiness, quality of life and Clinical Global Impression scales. Safety was comparable in both groups, as was drug consumption, suggesting a lack of any drug-seeking behaviour. Conclusion: Zolpidem administered 'as needed' is a feasible, well tolerated and promising additional option for the management of primary insomnia.