Cancer is one of the leading causes of death globally and its etiology needs to be understood to develop improved treatment options and minimize therapy resistance. This disease arises from multiple mutations that convert normal cells to cancer cells, and they continue to acquire mutations throughout tumor development. The second-most prominent source of mutagenesis across sequenced tumors comes from the activity of APOBEC3 cytidine deaminases. APOBEC3 enzymes are part of the innate immune response to combat foreign viral DNA and restrict transposon replication. APOBEC3s become dysregulated in cancer and their signature mutations (C to T or C to G at TCA or TCT in DNA) are prevalent in cervical, bladder, head and neck, lung, and breast cancers. There are seven different APOBEC3 proteins; to develop better cancer therapeutics, it is important to know which of the seven are contributing mutations in human cancer. Here, we study the contributions of APOBEC3 proteins in breast cancer and find that APOBEC3A (A3A) is likely the source of APOBEC signature mutations in breast cancer. Additionally, we provide correlative evidence that A3A is likely the main source of APOBEC signature mutations in head and neck, cervical, and bladder cancers as well. Within cancers, A3A-signature mutations are non-uniformly distributed, occurring not only with sequence specificity at TTCA motifs, but also showing DNA strand biases associated with replication direction and favoring hairpin forming sequences. This uneven distribution suggests that other factors besides sequence context contribute to APOBEC substrate preference. Using in vitro methods, we show that the APOBEC preference for hairpins over other ssDNA may be due to protein binding competition of longer ssDNA regions. In cancers experiencing replication stress, depletion of RPA that would otherwise compete with A3A for binding of ssDNA, results in enrichment of APOBEC signature mutations at the lagging strand of replication forks. We find that A3A activity is low on substrates representing transcription intermediates due to low substrate flexibility. Moreover, A3A activity is likely occluded by the presence of RNA polymerase on transcription intermediates. Together, these findings further explain why A3A signature mutations are non-uniformly distributed in cancer genomes.