The conventional definition and classifications of breast cancer delay are based on arbitrary empirical time cut-offs. In general, studies of cancer delay are based on these traditional definitions of patient and provider delay and are essentially atheoretical. If we aim to better understand delay, a reconsideration of its traditional conceptualisation and study methods is warranted. We propose a multidimensional model of breast cancer delay grounded in data from in-depth interviews with symptomatic patients and nested in the theory of illness behaviour. Our results show that delay prior to the first encounter with health services has to do with more than simply the patient as an individual, and delay posterior to this encounter is not due only to the health care providers. In fact, delay is a result of the interplay between the patient's socio-cultural context, individual characteristics that influence symptom interpretation and decision-making, interaction with the social network and types of support obtained, and aspects of the local health services. Future research on cancer delay should approach the problem integrally, taking into account the diverse dimensions involved.