This paper reports a field test of the prevailing antidune theory to explain the mechanisms of step-pool formation in steep mountain streams. The theory is supported by laboratory data but it has not received extensive field testing that would enable wide applicability to natural conditions. Data from the Santa Monica Mountains of California are used to test the antidune model where step wavelength and reconstructed hydraulic conditions are compared with the formation domain predicted by experimental and theoretical studies. Results show general agreement between field and laboratory data. Step wavelength and hydraulic requirements are consistent with those of the antidune process of formation, but results also suggest that true antidunes may be difficult to achieve in the steep headwaters where channels may be dominated by large roughness elements. These findings strengthen the link between theory and observation in clarifying how step-pools form; they contribute to the basis for elucidating why step-pools form in mountain streams. ¿ 1999 American Geophysical Union