Objective: The purpose of the present study was to develop a Danish adaptation of the Boston Naming Test (BNT) including a shortened 30-item version of the BNT for routine clinical use and two parallel 15-item versions for screening purposes. Method: The Danish adaptation of the BNT was based on ranking of items according to difficulty in a sample of older non-patients (n = 99). By selecting those items with the largest discrepancy in difficulty for non-patients compared to a mild Alzheimer’s disease (AD) sample (n = 53), the shortened versions of the BNT were developed. Using an overlapping cells approach preliminary education and age norms for older Danes were constructed. Results: The Danish adaptation of the BNT had adequate reliability and the short versions were all highly correlated with the full 60-item BNT (BNT-60). The sensitivity and specificity of the BNT-60 was.83 and.86, respectively. The shortened versions displayed some reduction in sensitivity (.70–.77) but similar or better specificity (.86–.91). Post-test probabilities of mild AD associated with performances along selected score ranges of the BNT were estimated. Likelihood ratios were presented that can be combined with information regarding the base rate of AD in any setting in order to assist in interpreting the clinical significance of a given performance. Conclusion: The short Danish versions of the BNT were highly correlated with the full BNT indicating that they measure the same construct. The Danish versions had acceptable diagnostic accuracy discriminating between mild AD and older non-patients.