The objective of the study was to analyze the prognostic factors of radiographic progression in a series of patients with early rheumatoid arthritis (RA) after 2 years of therapy with a structured algorithm using disease-modifying antirheumatic drugs (DMARDs) and very low doses of oral glucocorticoids. One hundred and five patients (81% female) with early RA (disease duration <2 years) treated with the same therapeutic protocol using gold salts and methotrexate in a step-up strategy, together with methylprednisolone (4 mg/day), were followed up for 2 years. The outcome variable was radiographic progression after 2 years of DMARD therapy using the modified Larsen method. Clinical, biological, immunogenetic, and radiographic data were analyzed at study entry and after 1 and 2 years of follow-up. Radiographic progression (increase of four or more units in the Larsen score) was observed in 32% of patients after 2 years of follow-up. The percentage of erosive disease increased from 18.3% at baseline to 28.9% at 12 months and 44.6% at 24 months, in spite of a significant improvement in disease activity. New erosions appeared in 33% of patients after 2 years. Several baseline parameters were associated with radiographic progression in the univariate analysis: shared epitope (SE) homozygozity, HLA-DRB*04 alleles, female gender, hemoglobin, erythrocyte sedimentation rate, and anticyclic citrullinated peptide antibodies (anti-CCP). In the multivariate analysis, female gender [odds ratio (OR) 5.5, 95% confidence interval (CI): 1.1-28.2, p = 0.04], DRB1*04 alleles (OR 3.1, 95% CI 1.1-9, p = 0.03) and, marginally, anti-CCP antibodies (OR 3.6, 95% CI 0.9-14.5, p = 0.06), were associated with progression. Female patients with both DRB1*04 alleles and anti-CCP antibodies showed the highest scores in radiographic progression. The presence, but not the titer, of anti-CCP antibodies predicted progression. The positive predictive value of the multivariate model for progression was only 53.9% whereas the negative predictive value was 80.3%. In a series of early RA patients treated with a structured algorithm using DMARDs and very low doses of glucocorticoids, radiographic progression was observed in one third of patients after 2 years. Female gender, DRB1*04 alleles (rather than the SE), and the presence of anti-CCP antibodies at baseline (independently of the titer) were the most important predictors of progression. The utility of these parameters in clinical practice is limited by their relatively low positive predictive value.