Food-chain length (FCL) is commonly used in ecological investigations to gain insight into how ecosystems function. Several studies have investigated the mechanisms underlying FCL patterns, but none has specifically examined the effect of temperature variability. In river ecosystems, water temperature variability can modify community structure, individuals' activity, and individuals' physiological rates, among other things. As such, we expected that it would negatively influence FCL. To test this prediction, we took advantage of a dataset comprising five streams, which mainly differ according to their temperature variability. At each stream, we (i) studied the species composition of macroinvertebrates and fish, and using nitrogen and carbon stable isotopes, (ii) estimated realized FCL, and (iii) examined food web structure. For macroinvertebrates, but not for fish, species composition differed among sites displaying low and high temperature variability. FCL was negatively influenced by temperature variability. Confirming this trend, we found a highly significant linear relationship between FCL and temperature variability using data from the literature. As for food web structure, the trophic position of filter-feeders/shredders may explain the FCL differences among sites. Our study gives additional support to the "dynamic stability" hypothesis and advances a step further by suggesting that temperature variability alone may reduce FCL.