In two 5-week trials, healthy college students were randomly assigned either to experimental or control groups. Participants in the experimental groups wrote about their affection for significant friends, relatives, and/or romantic partners for 20 minutes on three separate occasions; on the same schedule, those in the control groups wrote about innocuous topics. Total cholesterol was assessed via capillary blood at the beginning of the trials and again at the end. Participants in the experimental groups experienced statistically significant reductions in total cholesterol. Control participants in the first study experienced a significant increase during the same period, whereas those in the second study did not. Cholesterol changes were largely unmoderated by linguistic features of the writing produced in the intervention. Potential therapeutic implications are discussed.