A Cochrane review included 35 trials involving approximately 48 000 women, and resulting in 18 comparisons of 10 different IUDs. TCu380A was more effective in preventing pregnancy than MLCu375 (RD 1.70%, 95% CI 0.07% to 2.95% after 4 years of use). TCu380A was also more effective than MLCu250, TCu220 and TCu200. There tended to be fewer pregnancies with TCu380S compared to TCu380A after the first year of use, a difference which was statistically significant in the fourth year (RD -1.62%, 95% CI -3.00% to -0.24%). This occurred despite more expulsions with TCu380S (RD 3.50%, 95% CI 0.36% to 6.63% at 4 years). MLCu375 was no more effective than TCu220 at 1 year of use, or MLCu250 and NovaT up to 3 years. Compared to TCu380A or TCu380S, none of the IUDs showed any benefits in terms of bleeding or pain, or any of the other reasons for early discontinuation. None of the trials that reported events at insertion found one IUD easier to insert than another or caused less pain at insertion. There is no evidence that uterine perforation rates vary by type of device. There are minimal randomised data on IUD use in nulliparous women.