The temperature effect for the intensity of ,u-mesons at sea-level is investigated by taking 11-,u decays into account. The positive temperature coefficient is calculated by paying special attention to the difference in the attenuation mean free path of the N-component and the absorption mean free path of rr-mesons and to the atmospheric depth which gives main contributions to the coefficient. The magnitude of the coefficient is obtained as 0.05%/C and the depth of main contributions is found to be 200 mb. Influence of heavy mesons is qualitatively discussed § I. Introduction The atmospheric effect of cosmic rays has been interpreted in terms of the absorption of cosmic ray particles and of the spontaneous decay of p.-mesons. The former is correlated with the barometric effect, while the latter with the average height of the production of ,u-mesons. These effects were thoroughly analysed by Olberi> and there remains almost nothing added to his analysis, as far as the role of ,u-mesons is concerned. However, a possible interpretation of the positive temperature effect suggested by him seems to be an open question. Although Olbert tried to attribute the positive temperature effect to the difference in variations of two kinds of average heights introduced in his theory, these two are found to correlate with one another, so that Olbert's suggestion seems to be of little validity, as shown by Wada and Kudo2>. The positive temperature effect was first noticed by Forro3> for cosmic rays underground and was interpreted by Miyazima4> in terms of 1C·,tt decay. A detailed analysis of this effect showed that the upper limit of the positive temperature coefficient is about 0.5% j°C and is attained in the limit of high energies5>. The calculation of the coefficient was extended to lower energies by Barrett et al6> and their result predicted the magnitude of the coefficient appreciably smaller than 0.1% /°C for the bulk of the hard component observed at sea level. This had already been anticipated with a qualitative estimate by Duperier7> who observed a positive temperature coefficient of 0.1% j°C or larger for the hard component at sea level. This was found to be too high to reconcile with the theoretical prediction. On the other hand, several authorss-to> obtained lower values in rough agreement with the theoretical estimate. Such divergent experimental results are thought to be due partly to the varieties of ways of taking correlation with atmospheric temperatures. One can hardly say, at what