The contribution of intercepted cloud water to precipitation at windward and leeward cloud forest sites on the slopes of Haleakalā, Maui was assessed using two approaches. Canopy water balance estimates based on meteorological monitoring were compared with interpretations of fog screen measurements collected over a 2-year period at each location. The annual incident rainfall was 973 mm at the leeward site (Auwahi) and 2550 mm at the windward site (Waikamoi). At the leeward, dry forest site, throughfall was less than rainfall (87%), and, at the windward, wet forest site, throughfall exceeded rainfall (122%). Cloud water interception estimated from canopy water balance was 166 mm year−1 at Auwahi and 1212 mm year−1 at Waikamoi. Annual fog screen measurements of cloud water flux, corrected for wind-blown rainfall, were 132 and 3017 mm for the dry and wet sites respectively. Event totals of cloud water flux based on fog screen measurements were poorly correlated with event cloud water interception totals derived from the canopy water balance. Hence, the use of fixed planar fog screens to estimate cloud water interception is not recommended. At the wet windward site, cloud water interception made up 32% of the total precipitation, adding to the already substantial amount of rainfall. At the leeward dry site, cloud water interception was 15% of the total precipitation. Vegetation at the dry site, where trees are more exposed and isolated, was more efficient at intercepting the available cloud water than at the rainy site, but events were less frequent, shorter in duration and lower in intensity. A large proportion of intercepted cloud water, 74% and 83%, respectively for the two sites, was estimated to become throughfall, thus adding significantly to soil water at both sites. Published in 2010 by John Wiley & Sons, Ltd.