The structure of a concept generally labeled "teacher efficacy" is examined. A sample of 342 prospective and experienced teachers was administered an efficacy questionnaire adapted from the research of S. Gibson and M. H. Dembo (1984). Factor analytic procedures with varimax rotation were used to generate a 2-factor solution that accounted for 32 percent of the variance in scale scores. Contrary to previous research, these factors corresponded not to a personal versus teaching efficacy distinction, but instead to a simpler internal versus external distinction, similar to locus of control measures of causal attribution. Implications of these findings to past and future research on the construct are discussed. Five tables present analysis results, mostly factor loadings. One figure illustrates the discussion. (Author/SLD)