Objectives To evaluate the efficacy of the fetal heart diameter (HD) in predicting fetal hemoglobin (Hb) Bart disease at midpregnancy. Methods Video clips of fetal chest ultrasound examinations performed on fetuses at risk of Hb Bart disease at 18 to 22 weeks' gestation were randomly retrieved from our video clip database. The clips were replayed for fetal HD measurements, and the measured HDs were converted to z scores based on the z score model as a function of biparietal diameter. An HD z score greater than 2 or an actual HD value above the 95th percentile for gestational age was used as a cutoff in predicting Hb Bart disease. The best cutoff value of HD for identifying affected fetuses during midpregnancy was also evaluated by a receiver operating characteristic curve. Results A total of 90 video clips, including 37 affected and 53 unaffected fetuses, were measured. An HD z score greater than 2 had sensitivity of 94.6% (95% confidence interval [CI], 81.8%–99.3%) and specificity of 84.9% (95% CI, 72.4%–93.3%). An actual HD value above the 95th percentile for gestational age had sensitivity of 100% (95% CI, 90.5%–100%) and specificity of 69.8% (95% CI, 55.7%–81.7%). The best cutoff values of the HD z score and actual HD value were 2.27 (with sensitivity of 94.6% and specificity of 88.7%) and 18.15 mm (with sensitivity of 91.9% and specificity of 77.4%), respectively. Conclusions The fetal HD is highly effective in predicting fetal Hb Bart disease among fetuses at risk at midpregnancy. Both the HD z score and the actual value can be used for noninvasive prenatal screening of fetal cardiomegaly in clinical practice.