Due to optical processes in the atmosphere the spectral distribution of global, direct solar and diffuse radiation is different and depends on solar elevation, atmosphere transparency etc. Because of various optical properties of phytoelements, the spectral distribution of penetrated and reflected radiation differs from that of incoming radiation. Owing to radiative transfer processes, photosynthetically active radiation (PAR) and its proportion in integral solar radiation is not constant. The aim of the paper is to study theoretically these changes and to estimate systematic errors caused by deviation of the spectral sensitivity of pyranometers and PAR sensors from the spectral sensitivity of ideal sensors. The spectral distribution of radiation was estimated by combining different literature data and the spectral sensitivity of a typical pyranometer using available data for spectral transparency of glass domes and spectral reflectivity of black paints. Simultaneous use of energetic and quantum treatments of radiation in biogeophysics and subsequent existence of different definitions of PAR is a source of uncertainties and misunderstanding. The basic concepts, units and conversion factors between integral radiation and PAR, as well PAR efficiency for different kinds of solar radiation, and spectral corrections for some radiation sensors will be discussed below. Our calculations show that the conversion factor U PAR is 1–3% higher for penetrated radiation and about 6% higher for reflected radiation compared with U PAR Q=0.219 W s μmol −1 for global radiation under a cloudless sky. Due to the different spectral sensitivity of ideal energetic and ideal quantum sensors for PAR, the radiation measured through them differs 3–13% depending on the kind of radiation. Hence PAR quantum efficiency is 3–13% higher when PAR is defined as measured by the ideal energetic PAR sensor than when it is estimated by the ideal quantum PAR sensor. Their quantum efficiency depends essentially on the kind of radiation: for diffuse radiation it is increased by 13% but for reflected radiation it is decreased to about one fourth compared with global and direct solar radiation. For penetrated global radiation, quantum efficiency decreases rapidly with canopy depth, being 8–9 times lower at the bottom of a dense canopy than above it. For the LI-COR LI-190SA Quantum Sensor, systematic spectral errors do not exceed 1%. For the Kipp and Zonen PAR LITE sensor, errors are between 1 and 8% depending on the kind of radiation. For the LI-COR LI-200SA Pyranometer Sensor, systematic spectral errors for the global radiation reflected or penetrated by the canopy are high (20–40%), and this instrument cannot be recommended for measurements inside the canopy as is warned also by the manufacturers.