Recent interest in trade union activity has led to the development of econometric models of union membership growth. This paper examines the structural stability of two of the leading models—Ashenfelter-Pencavel's and Bain-Elsheikh's—each of which claimed to have captured the primary determinants of union growth in the twentieth century. The models were reestimated using revised, corrected, and extended membership data, and a nonlinear, maximum-likelihood procedure was employed to estimate the shift-point for each model. Contrary to previous studies, we found evidence of a break in the structure of each model. And unlike earlier work that hypothesized a World War II break-point, our estimated point was 1937–1938, most likely reflecting the impact of the Wagner Act.