The overall goal of this study is to use complexity science to gain a deeper understanding of the nonlinear day-to-day dynamics of intimate partner violence, with implications for clinical interventions. This report describes research methods for gathering information about partner violence in real time and assesses recruitment and retention, adherence to study protocol, data validity, and participant safety. Research assistants enrolled 200 women in moderately violent intimate relationships and asked them to report about their relationships every day for 12 weeks. Daily, participants telephoned an Interactive Voice Response (IVR) system and responded to 34 survey questions. They also completed baseline and end-of-study surveys and maintained telephone contact with the study team weekly. Forty-two participants completed qualitative end-of-study interviews to describe their relationships and the impact of the study on their lives. Of 200 enrollees, 145 women provided enough data for nonlinear analyses, averaging 63.5 daily reports of 84 possible. Participants submitted 9,201 daily reports, documenting partner's verbal or physical aggression on 39.4% of days, and their own aggression on 23.1%. Two women were withdrawn from the study for safety reasons; the remainder reported that study participation posed no additional threat. Eighty women sought assistance from community resources. Violence severity did not appear to change over the 12 weeks. The research team successfully and safely recruited and retained 145 women who provided valuable data for a study of complex dynamics of intimate partner violence.