Electronic devices such as mobile phones are quickly becoming a popular way to gather participant reports of everyday thoughts, feelings, and behaviors, including food cravings and intake. Electronic devices offer a number of advantages over alternative methods such as paper-and-pencil (PNP) assessment including automated prompts, on-the-fly data transmission, and participant familiarity with and ownership of the devices. However, only a handful of studies have systematically compared compliance between electronic and PNP methods of ecological momentary assessment (EMA), and none have examined eating specifically. Existing comparisons generally find greater compliance for electronic devices than PNP, but there is variability in the results across studies that may be accounted for by differences across research domains. Here, we compared the two EMA methods in an unexamined domain - eating - in terms of response rate and response latency, and their sensitivity to individual difference variables such as body mass index (BMI). Forty-four participants were randomly assigned to report on their food craving, food intake, and hunger four times each day for 2 weeks using either a PNP diary (N = 19) or text messaging (TXT; N = 25). Response rates were higher for TXT than PNP (96% vs. 70%) and latencies were faster (29 min vs. 79 min), and response rate and latency were less influenced by BMI in the TXT condition than in the PNP condition. These results support the feasibility of using text messaging for EMA in the eating domain, and more broadly highlight the ways that research domain-specific considerations (e.g., the importance of response latency in measuring short-lived food craving) interact with assessment modality during EMA.