AbstractBed load transport rate was measured in ten self‐formed small‐scale gravel braided streams developed in a laboratory flume at several different values of steady discharge and flume gradient. The streams are approximate Froude models of typical prototype braided streams but of no particular river. Slight viscous effects may be present in the models because particle Reynolds numbers are close to 70. Total bed load discharge was measured every fifteen minutes throughout each 60 hour run. In addition, 80 channel cross‐sections were measured in each run to establish the average channel geometry.Total bed load transport rate correlates well with total discharge and total stream power, although at a given stream power bed load discharge is greater when braiding is less intense and the width/depth ratio is lower. Analysis using unit stream power and cross‐section average bed shear stress reveals that the laboratory data conform to existing empirical bed load transport relationships. However, comparison with field data from gravel‐bed rivers shows discrepancies that may be due to differences in bed material size gradation and bed sediment structure.At constant discharge, wide fluctuations in bed load discharge occur with some regularity. Periods range from 2 to 10 hours in the models, which is equivalent to several tens of hours in a prototype. The presence of these long‐period fluctuations compounds the problems of field measurement of bed load in braided streams.