To assess the effect of a community-wide bicycle helmet campaign on helmet use, we observed 9827 children riding bicycles at sites in high-, middle-, and low-income census tracts in Seattle, Wash (intervention city), and Portland, Ore (control city); observations were made during 2-week intervals before and 4, 12, and 16 months after the campaign's start. Helmet use increased from 5.5% before the campaign to 15.7% afterward in Seattle and from 1.0% to 2.9% in Portland. Strong associations were found between helmet use and white compared with black or other race; riding geared vs nongeared bicycles; riding at playgrounds, in parks, or on bicycle paths vs on city streets; and riding with adults or other children compared with riding alone. The proportions of helmet wearers, adjusted for these variables, increased from 4.6% to 14.0% in Seattle and from 1.0% to 3.6% in Portland, a significantly greater increase in use in Seattle compared with Portland. We conclude that a community-wide bicycle helmet campaign can increase helmet use among children.