• This research analyzed how nuclear power plants implemented safety review innovations introduced by the Nuclear Regulatory Commission after the Three Mile Island accident. The findings suggested that nu clear power plants with relatively poor safety records tended to re spond in a rule-bound manner that perpetuated their poor safety per formance and that n"c1ear power plants whose safety records were relatively strong tended to retain their autonomy. a response that rein forced their strong safety performance.Innovations are ideas, formulas, or programs that the individuals in volved perceive as new (Beyer & Trice, 1978; Hill & Utterback, 1979; Rogers, 1982; Zaltman, Duncan, & Holbeck, 1973). The stages in their development have been the subject of much scholarly discussion and debate (Rothwell & Zegveld, 1985; Strebel, 1987). A common view is that implementation fol lows conception, proposal generation, and initiation, and that the factors that facilitate the former inhibit the latter (Duncan, 1976: 172; Wilson, 1963: 200). Rule-bound approaches. which involve central direction and highly programmed tasks, are supposed to promote implementation; that is. the number of routine tasks prescribed from above should increase as an organi zation moves toward implementation (Wilson, 1963: 198). Conception. pro posal generation. and initiation. on the other hand, require fewer controls and more autonomy. because diversity. openness, informality, and the abil ity to bring a variety of bases of information to bear on a problem need to be encouraged (Dunca~, 1976: 174). Duncan suggested that making the transi tion from conception, proposal generation, and initiation to implementation can be difficult and that the "ambidextrous" organization. adept at rna from stage to stage. is likely to be rare (1976: 167). Several authors (Burgelman. 1984; Kanter. 1985 Kanter. .1986 Lawrence &: DyeI'l 1982; Strebel, 1987) have dealt with the problem of a dominant corpo ", culture in established firms that is centered around rules that tend to ~,' innovation. They suggest that for innovation to occur. spin-offs. independent, task forces. and autonomous teams that simulate entrepreneurship w;necessary. Their analysis primarily applies to internally generated. oP~. tunity-driven innovation (Andrews. 1971; Bourgeois. 1984; Child, 1972),;i~ However. many innovations arise when an unanticipated external threat or', challenge occurs.' ,; The insight that crises. dissatisfaction. tension. and significant extemal~' stresses play an important role in bringing about innovations is a COIl1lIlOJl5f J one (Bateson. 1979; Crozier. 1964 Zander. 1977). To stimulate the introduction of the"'" new practices. disruptive events, which threaten a social system, may bifff needed. In fact. Terreberry (1971: 69) maintained that innovations are1 . a matter of external inducement.The problems that surface during the implementation of externally duced innovations. however, can thwart technological improvement distort the innovation process both directly...