An equipment for sampling flood water in rivers is devised and produced on a trial. This equipment is a series of bottles. Each bottle has a mouth which can be closed by a small float automatically after the water comes in. The bottles are previously set in a river, left in the flow during a flood and retrieved later. A flume test is done in order to know how well the bottle keeps the sampled water when it is exposed to a flow after the sampling. The result shows that most of the tested bottles can keep more than 90% of the sampled water for several hours. A field experiment is also carried out to confirm the possibility of practical use. The, water quality of sampled water is compared with that of the water which is sampled separately by other device. The both results agree with each other fairly well.