Researchers interested in working memory have debated whether it should be considered a single latent cognitive ability or a set of essentially independent latent abilities distinguished by domain-specific memory and/or processing resources. Simultaneously, researchers interested in cognitive aging have established that there are substantial differences in rates of change in various aspects of cognitive function with age. In general, so-called fluid measures of cognitive function including working memory decline at faster rates in later adulthood than so-called crystallized measures. Using an Internet working and short-term memory test battery completed by over 95,000 people aged 18-90, we used multiple-group confirmatory factor analysis to assess the extent to which working memory could be considered a single latent ability as well as how its common and unique variance components varied with age. Results indicated a single latent factor, but the tests did not reflect this factor consistently across age groups. Both individual test residual variances and factor intercepts showed different patterns of variation with age. We discuss the implications for understanding age differences in working memory function. (Contains 5 figures and 6 tables.)