Results of 2,500 inmate questionnaires and 1,200 guard questionnaires indicate that inmates who felt deprivation relative to the institution were more likely than those who experienced less institutional deprivation to support and participate in the riot, while inmates who experienced deprivation in interaction with fellow inmates were less, rather than more, likely to support and participate in the riot. Probably, inmates who feel that other inmates make their lives miserable are unlikely to act collectively with them. Because the collective behavior and social movement perspectives do not adequately explain inmate behavior in riots, an identification theory for inmate riots may be more appropriate. Prison riots are immensely variable phenomena. An identification theory would focus on the levels of inmates' identification with other prisoners, guards, officials, and with 'control communities' (political or social groups and institutions outside the prison). Social researchers should place less emphasis on predicting prison riots and more on the kinds of inmate behavior which will affect the kind of riot that will occur. Administrators should plan accordingly. Extensive tabular data, notes, a 74-item bibliography, and a summary of Michigan inmate and correctional officer attitudes are provided. The text quotes officer attitudes are provided. The text quotes extensively from inmate responses.