Accurate and detailed modeling of vegetation photosynthesis and ecosystem-atmosphere CO2 exchange, particularly for forests, requires data on the diffuse and global fluxes of photosynthetically active radiation (PAR) (400--700 nm). Such ecosystem process models typically employ PAR data in quantum units (photosynthetic photon flux density, Qp, ¿mol m-2 s-1), although PAR is often reported in energy units (irradiance, Rp, W m-2). Reliable conversion of diffuse PAR data from energy units (Rp,d, W m-2) to quantum units (Qp,d, ¿mol m-2 s-1) requires knowledge of how the conversion factor ¿*p,d (quanta-to-energy ratio for diffuse PAR at an arbitrary timescale, ¿mol J-1) varies across a broad range of sky conditions that includes variations in cloudiness, including broken clouds. This study employs spectral measurements of the diffuse irradiance of PAR to examine the influence of a diverse range of observed sky conditions on the spectral composition of Rp,d and the value of ¿*p,d at 1-minute and daily timescales. The results indicate that the enhanced contribution of green-to-red (550--700 nm) radiation to the spectral composition of Rp,d, induced by atmospheric scattering and diffuse reflectance by clouds, causes ¿*p,d at the daily timescale (¿'p,d) to increase nonlinearly by up to 7% relative to clean, cloudless sky conditions. The relation is explained well (R = 0.98) by the equation ¿'p,d = (4.5886 ¿ f'p,d)/(0.010773 + f'p,d), where f'p,d is equal to the diffuse fraction of the daily total irradiation of global PAR. These results may be useful for conversion of diffuse PAR data between energy and quantum units for environments where clouds are the dominant agent of atmospheric scattering. The observed quanta-to-energy ratio for global PAR at the daily timescale (¿'p,g), while not strictly constant, is relatively insensitive to atmospheric scattering, and a representative value of 4.56 ¿mol J-1 may be employed for a wide range of cloud conditions with little or no error.