STOPGAP is an open-source package for subtomogram averaging (STA), designed to provide users with fine control over each step of the process. It uses a real-space correlation-based approach, similar to other STA packages, but includes new algorithms for template matching, subtomogram alignment, and classification. The package addresses the challenges of cryo-electron tomography (cryo-ET), where overlapping molecular structures complicate structural analysis. STOPGAP improves the performance of template matching and subtomogram alignment by accounting for the missing wedge, a region of Fourier space that is not sampled due to limited tilt angles. It also includes algorithms for subtomogram classification using multi-reference alignment (MRA), which allows for the assessment of classification reproducibility through stochastic methods. STOPGAP is implemented in MATLAB and includes a main executable and a set of functions and scripts. It supports parallel processing using MPI or SLURM. The package includes a parser to check parameters and generate formatted files for various tasks. The missing-wedge model is implemented as a series of Fourier slices, with local CTF filters used for subtomogram alignment, averaging, and classification. Constrained cross correlations (CCCs) are calculated using a 3D version of the fast local correlation function (FLCF), with reference and search maps filtered to remove artifacts. Template matching involves splitting tomograms into subvolumes (tiles) for parallel processing, rotating templates through a set of orientations, and calculating CCCs. A noise-correlation approach is used to reweight the main score map, improving the accuracy of template matching. Subtomogram extraction is performed by cropping subvolumes according to positions stored in a motive list. Subtomogram alignment and averaging are performed as 'halfsets', with the motive list split into two and aligned independently. The final alignment results in a figure-of-merit weighted sum of the halfmaps. Classification by MRA involves generating de novo references from the data set and aligning the full data set against multiple references. The process includes simulated annealing and stochastic hill-climbing to avoid local optima and improve classification reproducibility. The package has been tested on 80S ribosome and HIV s-CANC data, demonstrating improved resolution and accuracy in subtomogram alignment and classification. The missing-wedge model in STOPGAP improves the quality and interpretability of averaged maps by accounting for anisotropic sampling and amplitude modulations. The package is designed to be a technical resource for users and for further community-driven software development.STOPGAP is an open-source package for subtomogram averaging (STA), designed to provide users with fine control over each step of the process. It uses a real-space correlation-based approach, similar to other STA packages, but includes new algorithms for template matching, subtomogram alignment, and classification. The package addresses the challenges of cryo-electron tomography (cryo-ET), where overlapping molecular structures complicate structural analysis. STOPGAP improves the performance of template matching and subtomogram alignment by accounting for the missing wedge, a region of Fourier space that is not sampled due to limited tilt angles. It also includes algorithms for subtomogram classification using multi-reference alignment (MRA), which allows for the assessment of classification reproducibility through stochastic methods. STOPGAP is implemented in MATLAB and includes a main executable and a set of functions and scripts. It supports parallel processing using MPI or SLURM. The package includes a parser to check parameters and generate formatted files for various tasks. The missing-wedge model is implemented as a series of Fourier slices, with local CTF filters used for subtomogram alignment, averaging, and classification. Constrained cross correlations (CCCs) are calculated using a 3D version of the fast local correlation function (FLCF), with reference and search maps filtered to remove artifacts. Template matching involves splitting tomograms into subvolumes (tiles) for parallel processing, rotating templates through a set of orientations, and calculating CCCs. A noise-correlation approach is used to reweight the main score map, improving the accuracy of template matching. Subtomogram extraction is performed by cropping subvolumes according to positions stored in a motive list. Subtomogram alignment and averaging are performed as 'halfsets', with the motive list split into two and aligned independently. The final alignment results in a figure-of-merit weighted sum of the halfmaps. Classification by MRA involves generating de novo references from the data set and aligning the full data set against multiple references. The process includes simulated annealing and stochastic hill-climbing to avoid local optima and improve classification reproducibility. The package has been tested on 80S ribosome and HIV s-CANC data, demonstrating improved resolution and accuracy in subtomogram alignment and classification. The missing-wedge model in STOPGAP improves the quality and interpretability of averaged maps by accounting for anisotropic sampling and amplitude modulations. The package is designed to be a technical resource for users and for further community-driven software development.