Aim To assess whether altitude changes in the distribution of butterflies during the second half of the 20th century are consistent with climate warming scenarios. Location The Czech Republic. Methods Distributional data were taken from a recent butterfly distribution atlas, which maps all Czech butterflies using a grid of 10′ longitude to 6′ latitude, equivalent to about 11.1 × 12 km. Cell altitude was used as an independent variable, and altitudinal ranges of individual species (less migrants, extinct species, recent arrivals and extremely rare species) in 1950–80 vs. 1995–2001 and in 1950–80, 1981–94, 1995–2001 were compared using U‐tests and linear regressions. Results Of 117 (U‐tests) and 119 (regressions) species, we found significant uphill increases in 15 and 12 species, respectively. The two groups were nested; none (U‐test) and one (regression) species showed a significant altitudinal decrease. Binomial tests of frequencies of signs of the U‐tests and regression coefficients, including nonsignificant ones, also showed that the increases prevailed. The mean and median of the significant shifts were 60 and 90 m, respectively, and the maximum shift per species was 148 m. The recording effort in individual time periods was not biased with respect to altitude. Main conclusion Altitude shifts in the distribution of Czech butterflies are already detectable on the coarse scales of standard distribution maps. The increasing species do not show any consistent pattern in habitat affiliations, conservation status and mountain vs. nonmountain distribution, which renders climatic explanation as the most likely cause of the distributional shifts.