Purpose: This study aims to describe the distribution of ultra-processed food and drink products (UPP) consumption according to sociodemographic characteristics in adults from southern Brazil, and to investigate which are the most-consumed UPP subtypes in the different social strata. Methods: Cross-sectional analysis of the second wave of a population-based cohort of 1720 adults. The usual caloric intake and the caloric contribution of UPP to total energy intake (%CTEI) were estimated by the application of two 24-h dietary recalls (adjusted by intra- and inter-individual variability). Data were analyzed according to gender, age, marital status, schooling, and family income. Linear regression models were used to estimate the adjusted means. Results: Consumption data were obtained from 1206 adults (70.1% of the original cohort). Mean UPP consumption was higher in males than females (829.6 kcal vs 694.3 kcal, p value < 0.001), but the %CTEI from UPP increased in females (34.7% vs 39.3%, p value < 0.001), even after adjusting for sociodemographic variables. In the full model, which included all sociodemographic variables, %CTEI from UPP was inversely associated with age (difference between extreme categories 7.1 pp., 95 CI% 7.7–6.5) and directly associated with schooling (difference between extreme categories 6.3 pp., 95 CI% 5.5–7.1). The subtypes of UPP that contributed most to the observed differences were processed breads, fast food, and ultra-processed pies and sweets. Conclusions: UPP account for a third of the calories normally consumed, with women, young people, and better educated individuals being the most vulnerable groups. These results can help when planning public policies to reduce UPP consumption.