Transitive inference (TI) is a form of deductive reasoning requiring the ability to infer a serial relationship between two items never explicitly compared. Two studies were conducted to examine age differences in TI reasoning in healthy younger (aged 18-35 years) and older (aged 60+ years) individuals. TI task complexity was manipulated using positive and negative phrasing (e.g., "taller" vs. "not taller"), by increasing the memory load (removing premise sentences while the inference is generated relative to having them remain in view), and by increasing number of premise terms (three, four, and five terms). In both studies, younger adults had higher accuracy than older adults, and the use of five-term premises reduced all participants' accuracy relative to three- and four-terms. Performance for both age groups was better with premises present compared to premises absent or negative phrasing. Reaction time data revealed similar patterns for both age groups, and although premises absent yielded faster responding compared to premises present or negative phrasing, the accuracy data suggested that the TI task was particularly difficult if premises did not remain in view. The relationship between TI performance and neuropsychological measures of executive function, reasoning, and working memory was examined in the second study. In older adults, performance was correlated with verbal reasoning, executive function, and linguistic skill, whereas in younger adults, performance was correlated with working memory and processing speed. Thus, to solve TI tasks, older individuals may rely more on stable resources like verbal reasoning and linguistic skill than on working memory.