Very few earthquake risk studies exist for cities on the Indian subcontinent. The few studies that do exist typically focus on intensity as the parameter to describe the expected ground motion during an earthquake and on damage observations to represent building vulnerability. In contrast to these empirical studies, analytical loss computations, which are based on capacity spectrum methods (CSM), have recently become popular and are gaining wide acceptance. Analytical damage and loss computations have been conducted for the test bed Dehradun, a city of 500,000 inhabitants in the foothills of the Himalayas (northern India), and then compared with loss estimates from empirical studies recently performed for the city. The study illustrates the problems associated with trying to generate intensity-compatible ground motion estimates and comparing the damage and loss estimates of both approaches.