Currently, information on the psychometric properties of the Medical outcomes study-social support survey (MOS-SSS) for patients with chronic disease in primary health care, suggests problems in the dimensionality, specifically predominant unidimensionality in a multidimensional measure. The aim of this study was to determine the internal structure (dimensionality, measurement invariance and reliability) and association with other variables.A total of 470 patients with chronic disease from a Family Medicine Unit at the Instituto Mexicano del Seguro Social, IMSS, with a mean age of 51.51 years were included. Participants responded to the Questionnaire of Sociodemographic Variables (Q-SV), SF-36 Health-Related Quality of Life Scale-version 1.1, and MOS-SSS.Non-parametric (Mokken scaling analysis) and parametric (confirmatory factor analysis) analyses indicated unidimensionality, and three-factor model was not representative. A new 8-item version (MOS-S) was developed, where measurement invariance, equivalence with the long version, reliability, and relationship with the SF-36 were satisfactory.The MOS-SSS scale is unidimensional, and the shortened version yields valid and reliable scores for measuring social support in patients with chronic disease at the primary health care.