A reaction processor is provided with a reaction processing vessel in which a channel is formed, a liquid feeding system, a temperature control system for providing a high temperature region and a low temperature region to the channel, and a fluorescence detector for detecting the sample passing through a fluorescence detection region of the channel, and a CPU for controlling the liquid feeding system based on a signal that is detected. A target stop position X[L]0(n+1) of the sample in the low temperature region in an (n+1)th cycle is corrected from a target stop position X[L]0(n) of the sample in the low temperature region in the nth cycle based on the result of stopping control on the sample in the nth cycle.