Recent data indicate that domestic violence affects a considerable proportion of welfare recipients and other poor women. The optional provisions for battered welfare recipients in the welfare reforms of 1996 (known as the Family Violence Option) were based on the assumption that going to work aggravates battering and PTSD symptoms. One way to measure the effects of work on domestic abuse is to ask women whether their working changed the onset, frequency, or severity of abuse. This research contributes to knowledge about poverty and abuse by analyzing the subjective assessments welfare recipients gave of the effects their working had on abuse and the symptoms of traumatic stress. During May to June 1998 and June 2001, researchers conducted structured interviews with welfare recipients enrolled in “work first” (job search, employment skills, and paid work experience) activities in Allegheny County, PA. Enrollees were eligible if they were women, receiving Temporary Aid to Needy Families (the cash benefit for poor mothers and children), not pregnant, and at least 18 years old. In the 2 cohorts (1998 and 2001), 162 women participated in the interviews. The two studies shared a common set of measures of physical violence, work-related interference and sabotage, and symptoms related to PTSD. The findings show that, for many women, battering was aggravated by going to work. For the majority of those that reported either physical abuse or work-related interference and control, going to work either precipitated or aggravated the abuse, or seemed to have no effect. Only a minority reported that working made the abuse slacken or stop. In addition, work interruption and the pay penalty were generally worse for those women that reported that working precipitated or aggravated battering or PTSD symptoms. Between one in three and four in five respondents with PTSD symptoms reported that working brought some relief. These findings suggest that it is important for researchers to differentiate among measures of battering and its consequences. 4 tables, 26 references