No data exist about the possibility that vertebral fracture in PMR patients could be independent of steroid therapy. For this reason, we aimed to investigate this topic by a case cohort study with a 1-year follow-up for each patient. We selected ten consecutive patients who experienced vertebral fractures (VF-group) during the first month of 1-year follow-up period and without any other significant associated condition. As a control group we studied ten control patients, without vertebral fractures and with a follow-up of 1 year, randomly selected among a larger group of patients affected by polymyalgia rheumatica. The following data were analysed: eritrosedimention rate (ESR), visual analogical scale score (VAS), methyprednisolone daily dosage. Each patient had been monthly evaluated by the aforementioned clinical and laboratoristic parameters during the 1-year follow-up period. The VF-group resulted with a higher and statistically significant median corticosteroid 12-month total dosage [mean 3,480 mg (95%CI 2,805-3,030) vs. 2,760 mg (2,666.25-3,247.5)]. The VF-group had statistically significant higher ESR and VAS AUC when compared to control group (median ESR AUC, 484.75 vs. 288.25; P = 0.0001; median VAS AUC, 70.75 vs. 43.5 P < 0.0001); ESR at the baseline (cut-off >80 mm) showed a specificity of 90% (95%CI 56-100) and sensitivity of 70% (95%CI 35-93). VAS difference from first to second month (cut-off <or=3) showed a specificity of 90% (95%CI 56-100) and sensitivity of 80% (95% CI 44-97). Our results point out that vertebral fracture might be predicted from commonly used laboratory (ESR) and clinical (VAS) findings.