Abstract Background Hip fractures are common in the elderly and have a high risk of early mortality. Identification of patients at high risk of early mortality could contribute to enhanced quality of care. A simple scoring system is essential for preoperative identification of patients at high risk of early mortality in clinical practice. Of risk models published, The Nottingham Hip Fracture Score (NHFS) shows the most promising results so far. However, there is still room for improvement. Methods A cohort study including 850 patients was conducted over a period of 5,5 yr. The NHFS was adjusted for cognitive impairment (NHFS-a) and tested. Patients who died within 30 days following hip fracture surgery (early mortality group) were compared to survivors. Independent risk factors for early mortality were assessed. A new hip fracture score for frail elderly was developed: the Almelo Hip Fracture Score (AHFS). The NHFS-a and the AHFS were compared for accuracy and predictive validity. Results Sixty-four (7.5%) patients died within 30 days following hip fracture surgery. The AHFS predicts the risk of early mortality better than the NHFS-a (p < 0.05). Using cut-off points of AHFS ≤ 9 and AHFS ≥ 13, patients could be divided into a low, medium or high risk group. The area under the curve improved with the AHFS compared to the NHFS-a (0.82 versus 0.72). The likelihood ratio test reveals a significantly better fit of the AHFS in comparison with the NHFS-a (p < 0.001). Conclusions The AHFS can identify frail elderly at high risk of early mortality following hip fracture surgery accurately. With the AHFS, the patient can be classified into the low, medium or high risk group, which contributes to enhanced quality of care in clinical practice.