The aim of this study was to determine whether attributional style is a stable pattern in somatizers, to analyse the sociodemographic and psychopathological characteristics that can modify it, and to study the relationship between attributional pattern and outcome. A total of 147 somatizers and 46 psychologizers from a representative sample (n = 1559) of primary care patients in Zaragoza, Spain were followed up for 1 year. Attribution of somatic symptoms was a stable construct in somatizers. Patients who modified attribution were younger (by 15 years on average), tended to be without a partner, and had a shorter illness duration (by 20 months on average) than those who maintained it. Attribution showed no correlation with outcome at the 1-year follow-up.