A broadleaf mixed forest diversified through partial tree thinning was studied to identify expedient sampling and data analysis procedures to capture the heterogeneous within-canopy downward distribution of instantaneous global photosynthetic photon flux (PPF); to extract foliage structural properties from the acquired light values; and to compute statistics descriptive of the within-canopy light and leaf layer distributions. We sampled PPF at 1-m intervals along vertical gradients using a helium-filled balloon as a platform for a light sensor. A random method was used to identify the forest floor locations for the within-canopy balloon ascents. About 400 PPF measurements were recorded per vertical transect. For each PPF value, we computed, by inversion of the Monsi-Saeki model, the number of leaf strata cumulated along the sunbeam direction from the position where the light was measured. Variability in PPF and leaf layer at different vegetation scales was computed by non-parametric statistics. The methods were evaluated as appropriate for intra-canopy PPF sampling, particularly in an undisturbed canopy. The minimum number of vertical PPF profiles required to capture the within-canopy PPF variability was 9-10 (equivalent to about 4000 measurements). The reliability and sensitivity of the inversion of the Monsi-Saeki method were sufficient to capture the canopy structural differences between undisturbed and partially thinned forests. The proposed PPF canopy sampling and data analysis procedures provide a fast, reliable and inexpensive way to characterize tree crown structure, and to predict plant growth and forest dynamics and could be applied whenever vegetation absorbed radiation is a main driving force for forest canopy processes. The experimental light attenuation data and the extracted canopy leaf layer numbers could serve to corroborate canopy mechanistic models of radiative transfer and net primary production.