Adult zebrafish are frequently described to be able to "completely" regenerate the heart. Yet, the extent to which cardiomyocytes lost to injury are replaced is unknown, since only indirect evidence for cardiomyocyte proliferation exists. We established stereological methods to quantify the number of cardiomyocytes at several time-points post cryoinjury. Intriguingly, after cryoinjuries that killed about 1/3 of the ventricular cardiomyocytes, pre-injury cardiomyocyte numbers were restored already within 30 days. Yet, many hearts retained small residual scars, and a subset of cardiomyocytes bordering these fibrotic areas remained smaller, lacked differentiated sarcomeric structures, and displayed defective calcium signaling. Thus, a subset of regenerated cardiomyocytes failed to fully mature. While lineage-tracing experiments have shown that regenerating cardiomyocytes are derived from differentiated cardiomyocytes, technical limitations have previously made it impossible to test whether cardiomyocyte trans-differentiation contributes to regeneration of non-myocyte cell lineages. Using Cre responder lines that are expressed in all major cell types of the heart, we found no evidence for cardiomyocyte transdifferentiation into endothelial, epicardial, fibroblast or immune cell lineages. Overall, our results imply a refined answer to the question whether zebrafish can completely regenerate the heart: in response to cryoinjury, preinjury cardiomyocyte numbers are indeed completely regenerated, while restoration of cardiomyocyte differentiation and function, as well as resorption of scar tissue, is less robustly achieved.