The proportion of prone sleeping among sudden infant death syndrome (SIDS) victims and infants in general, and the rate of SIDS were prospectively studied in the county of Hordaland, Norway, three years before (1987-89) and three years after (1990-92) a campaign to discourage prone sleeping. Before the campaign, 64% of random reference infants were put prone versus 8% after (p < 0.0001). Concurrently, the rate of SIDS decreased from 3.5 to 1.6 per 1000 live births (63 infants before and 30 after the campaign, p = 0.0002). Prone sleeping was not considered a statistically significant risk factor for SIDS before (OR 2.0, 95% CI 0.8-4.5), but was highly significant (OR 11.3, 95% CI 3.6-36.5) after the campaign. Prone sleeping is an important risk factor for SIDS, but the association may be missed in epidemiological studies if prone is the predominant sleeping position. Behaviour with regard to sleeping position may be changed rapidly by means of a simple campaign.