Many spatial analysis techniques rely on the ability to geocode individual locations based on addresses or other descriptive information. The quality of geocoding and its effect on spatial analysis have received some attention in the literature, in particular in the field of health. This article reviews the foundation of geocoding and presents a framework for evaluating geocoding quality. Errors introduced by street gecoding include incompleteness, positional error, and incorrect assignment to geographic units. A review of empirical studies suggests that these errors are neither small nor random in nature and that substantial bias may be introduced in spatial analysis that employs the results of geocoding. Several alternatives have also emerged, including the use of address points and parcels, and these are gradually becoming more widely used. Several areas for future research on geocoding have been identified: (i) refinements of address data models to incorporate complex addressing situations; (ii) development of error propagation techniques to determine the level of geocoding quality required for a particular analysis scenario; (iii) development of measures of reliability for geocoding results; (iv) comparative analysis of geocoding quality across different jurisdictions; and (v) validation of online geocoding services and volunteered geographic information.