Data from the Australian Bureau of Statistics three cross-sectional National Health Surveys, conducted in 1989-1990, 1995-1996 and 2000, were used to examine long-term population responses to the 1996 physical activity (PA) recommendations. Past 2-week recall of the number of times and total minutes spent walking for recreation or sport, and doing moderate and vigorous exercise, were averaged for 1 week to create a measure of 'sufficiently active': >or=5 weekly sessions of >or=150 min/week of at least moderate-intensity activity. With the exception of adults aged>60 years, from 1989 to 2000, there were increases in the prevalence of walking (+8%) and in moderate-intensity exercise (+4%), resulting in a significant decline (-6%) in the proportion of those reporting no leisure-time exercise. Participation in vigorous exercise remained unchanged. There was a 'population shift' in median minutes and sessions of walking from 1995 to 2000, but the prevalence of regular walking (>or=5 times and >or=150 min/week) remained unchanged. A small but significant increase (+2%) in the proportion of adults who met the PA recommendations was noted mainly for men under 60 years and for women aged 45-59 years. The proportion meeting the PA recommendations declined significantly in the most socioeconomically disadvantaged area. These data provide evidence contrary to the observed declines in PA documented by the Active Australia surveillance in the late 1990s. The population may have responded to the 1996 concept of moderate-intensity exercise mainly by walking, but sustainable promotional efforts will be required to build on these gains in the forthcoming decades.