The following criteria represent changes or additions to the original protocol for this review:

1) The original protocol specified an intended limit to the interventions to be considered, i.e. the four major NSAIDs prescribed for dysmenorrhoea (naproxen, ibuprofen, indomethacin, mefenamic acid) and two commonly used non‐opioid analgesics (aspirin, paracetamol). However while searching was being carried out it became clear that a much wider group of NSAIDs are now commonly used and also that aspirin is itself commonly considered a NSAID, so it was decided to make the review more comprehensive in its evaluation by considering all types of NSAIDs, including aspirin, versus other NSAIDs, paracetamol or placebo.

2) It was decided to include only double blinded trials, and to exclude trials which analysed fewer than 80% of randomised participants, on grounds of quality.

3) Trials with participants who used intrauterine contraception (IUCD) were excluded unless the results for this group were analysed separately, on the grounds that dysmenorrhoea related to IUCD use is secondary rather than primary (Ylikorkala 1978).

4) Where trials compared differing doses of the same NSAID against placebo or another drug, the results were pooled where feasible, or (in the case of crossover trials where results could not be pooled) the highest recommended dose was used for the purpose of comparison. See Characteristics of Included Studies for details.

5) Cyclooxygenase‐2 inhibitors (Cox 2 inhibitors), such as praecox and celecoxib, were excluded as they will be considered in a separate review.