The Zn diffusion into InP and In0.57Ga0.43As was studied. A comparison of three diffusion methods is presented, i.e., boat diffusion, diffusion from As‐ or P‐doped spin‐on films, and diffusion from In‐doped spin‐on films. The diffusion into samples made by different growth methods was investigated, i.e., bulk grown wafers, liquid‐phase epitaxial films (LPE), and metalorganic vapor‐phase epitaxial films (MOVPE). The in‐depth profiles found are from the double diffusion front type, p+ ‐p−‐n. The p+ ‐p− junction position depends on the diffusion method but not significantly on the growth method of the sample, while the p−‐n junction depends on both. For example, with As or P in the spin‐on film, a shallow p+ ‐p− junction and a deep p−‐n junction appear. With In in the spin‐on film, a deep p+ ‐p− junction with a negligibly small p− region is found. Diffusion into bulk and MOVPE samples usually yields deeper p−‐n junctions than into LPE samples. Owing to the fact that in the presented experiments the amount of P, As, In, Ga, or the respective vacancies VP, VAs, VIn, VGa differ, we are able to check which of the existing diffusion models are applicable here. We propose an independent trapping of the mobile interstitially diffusing Zni by two immobile vacancy centers, i.e., Zn on VIn or VGa in the p+ region and Zn on VAsZnVAs or VPZnVP in the p− region.