Electronic registries for clinical trials present an opportunity for miscommunication of benefits and risks to potential subjects. We explored early phase clinical trials registered on the National Library of Medicine’s ClinicalTrials.gov for inappropriate language patterns (terms and statements) that could convey therapeutic intent. We sampled early-phase heart disease clinical trial records involving stem cells registered on Clincaltrials.gov between April 24, 2004, and September 18, 2008, and randomly selected heart disease drug trials from the same period for comparison. Trial sponsorship and location were coded. Textual data were extracted from titles, purpose, outcome measures, and detailed description statements and coded for readability and the presence of therapeutic language. Content and multiple correspondence analyses were employed to explore language patterns suggestive of therapeutic intent. We found statistically significant differences in therapeutic language use between stem cell (n = 72, median statement frequency 3.5, IQR 1–7) and drug trial (n = 72 median statement frequency 1, IQR 0–3.5) records; Mann–Whitney U = 1808.5,p = .001. A correspondence plot, accounting for 65% of data variability, suggests correspondences among patterns of therapeutic language, regional location, and type of sponsorship between the two record types. Individual-, institute-, and government-sponsored stem-cell trials showed a tendency toward high frequency use of therapeutic statements. This study, the first to explore the potential for therapeutic miscommunication in electronic registries, suggests the need for greater scrutiny of language used in registered trials. Recommendations are offered to improve the integrity of records submitted to ClinicalTrials.gov.