Spatial microheterogeneity (∼100 mm scale) of atmospheric deposition was studied to re-examine the conventional methodology for P input estimates. From September to December 2000, nine replicated bulk samples were obtained after each rain event at Ashiu Experimental Forest, and analyzed for P, Na and Mg. Na and Mg contents were homogeneous among samples (mean CV = 8.9 and 16%, respectively), showing that almost all of the elements were derived from sea salt. On the other hand, P deposition had remarkable heterogeneity (mean CV = 101%). The contrast between P and Na/Mg data, microscopic observation of samples, and the high correlation between P deposition and air temperature suggest that local sources, especially insects and pollen, are the origin of the large heterogeneity. Data screening methods to obtain a clean dataset showed that local sources add more P (> 1.21 μmol m-2 day-1) than the sources transported over a long distance (<0.15 μmol m-2 day-1). It depends on the target ecosystem whether the local sources should be considered as contamination or part of the true atmospheric input. In any case, large heterogeneity should be treated adequately in P deposition measurement. Conventional methodology, which extrapolates poorly replicated data to the whole area of interest, can be the cause of a serious error. To accurately gauge atmospheric P input, one must carefully choose methodologies that are suited to the specific ecosystem and the particular P sources.