Tissue-engineered in vitro models are an essential tool in biomedical research. Tissue geometry is a key determinant of function, but controlling the geometry of microscale tissues remains challenging. Additive manufacturing approaches have emerged as a promising means for rapid and iterative changes in the geometry of microdevices. However, it has been shown that poly(dimethylsiloxane) (PDMS) cross-linking is often inhibited at the interface of materials printed with stereolithography. While approaches to replica mold stereolithographic three-dimensional (3D) prints have been described, these methods are inconsistent and often lead to print destruction when unsuccessful. Additionally, 3D-printed materials often leach toxic chemicals into directly molded PDMS. Here, we developed a double molding approach that allows precise replication of high-resolution stereolithographic prints into poly(dimethylsiloxane) (PDMS) elastomer, facilitating rapid design iterations and highly parallelized sample production. Inspired by lost wax casting, we used hydrogels as intermediary molds to transfer high-resolution features from high-resolution 3D prints into PDMS, while previously published work focused on enabling direct molding of PDMS onto 3D prints through the use of coatings and post-crosslinking treatments of the 3D print itself. Hydrogel mechanical properties, including cross-link density, predict replication fidelity. We demonstrate the ability of this approach to replicate a variety of shapes that would be impossible to create using photolithography techniques traditionally used to create engineered tissue designs. This method also enabled the replication of 3D-printed features into PDMS that would not be possible with direct molding as the stiffness of these materials leads to material fracture when unmolding, while the increased toughness in the hydrogels can elastically deform around complex features and maintain replication fidelity. Finally, we highlight the ability of this method to minimize the potential for toxic materials to transfer from the original 3D print into the PDMS replica, enhancing its use for biological applications. This minimization of the transfer of toxic materials has not been reported in other previously reported methods describing replication of 3D prints into PDMS, and we demonstrate its use through the creation of stem cell-derived microheart muscles. This method can also be used in future studies to understand the effects of geometry on engineered tissues and their constitutive cells.