The relationship between canopy structure and light transmission to the forest floor is of particular interest for studying the effects of succession, timber harvest, and silviculture prescriptions on understory plants and trees. Indirect measurements of leaf area index (LAI) estimated using gap fraction analysis with linear and hemispheric sensors have been commonly used to assess radiation interception by the canopy, although the two methods often yield inconsistent results. We compared simultaneously obtained measurements of LAI from a linear ceptometer and digital hemispheric photography in 21 forest stands on Prince of Wales Island, Alaska. We assessed the relationship between these estimates and allometric LAI based on tree diameter at breast height (LAIDBH). LAI values measured at 79 stations in thinned, un-thinned controls, old-growth and clearcut stands were highly correlated between the linear sensor (AccuPAR) and hemispheric photography, but the latter was more negatively biased compared to LAIDBH. In contrast, AccuPAR values were more similar to LAIDBH in all stands with basal area less than 30 m(2)ha(-1). Values produced by integrating hemispheric photographs over the zenith angles 0-75° (Ring 5) were highly correlated with those integrated over the zenith angles 0-60° (Ring 4), although the discrepancies between the two measures were significant. On average, the AccuPAR estimates were 53% higher than those derived from Ring 5, with most of the differences in closed canopy stands (unthinned controls and old-growth) and less so in clearcuts. Following typical patterns of canopy closure, AccuPAR LAI values were higher in dense control stands than in old-growth, whereas the opposite was derived from Ring 5 analyses. Based on our results we advocate the preferential use of linear sensors where canopy openness is low, canopies are tall, and leaf distributions are clumped and angles are variable, as is common in the conifer forests of coastal Alaska.