Secondary forests occupy a larger area than old-growth rain forests in many tropical regions but their hydrological functioning is still poorly understood. In particular, little is known about the various components of evapotranspiration in these possibly vigorously regenerating forests. This paper reports on a comparison of measured and modeled canopy interception losses (I) from a semi-mature (ca. 20 years) and a young (5–7 years) secondary forest in the lower montane rain forest zone of eastern Madagascar. Measurements of gross rainfall (P), throughfall (Tf), and stemflow (Sf) were made in both forests for one year (October 2014–September 2015) and the revised analytical model of Gash et al. (1995) was tested for the first time in a tropical secondary forest setting. Overall measured Tf, Sf and derived I in the semi mature forest were 71.0%, 1.7% and 27.3% of incident P, respectively. Corresponding values for the young forest were 75.8%, 6.2% and 18.0%. The high Sf for the young forest reflects the strongly upward thrusting habit of the branches of the dominant species (Psiadia altissima), which favours funneling of P. The value of I for the semi-mature forest is similar to values reported for old-growth tropical lower montane rain
forests elsewhere but I for the younger forest is higher than reported for similarly aged tropical lowland forests. These findings can be explained largely by the prevailing low rainfall intensities and the frequent occurrence of small rainfall events. The revised analytical model was able to reproduce measured cumulative I at the two sites accurately and succeeded in capturing the variability in I associated with the seasonal variability in rainfall intensity, provided that Tf-based values for the average wet-canopy evaporation rates were used instead of values derived with the Penman-Monteith equation.