This thesis investigates how breaking apart selection interference (‘Hill-Robertson’ effects) that arises between linked loci can select for higher levels of recombination. Specifically, it mainly studies how the presence of both advantageous and deleterious mutation affects selection for recombination. These evolutionary advantages are subsequently investigated with regards to sex resisting asexual invasion in a subdivided population. i) KEIGHTLEY and OTTO (2006) showed a strong advantage to recombination in breaking apart selection interference, if it acts across multiple, linked loci subject to recurrent deleterious mutation. Their model is modified to consider selection acting on recombination if a small proportion of mutations are advantageous. This leads to a greater increase in selection acting on a recombination modifier, compared to cases where only deleterious mutations are present. ii) Branching-process methods are developed to quantify how likely it is that a deleterious mutant hitchhikes with a selective sweep, and how recombination between the two loci affects this process. This is compared to the neutral hitchhiking model, to determine how levels of linked neutral diversity would differ between the two scenarios. A simple application with regards to human genetic data is provided. iii) Population subdivision can maintain costly sex, as a consequence of restricted gene flow slowing the spread of invading asexuals, which leads to an excessive accumulation of deleterious alleles. However, previous work did not quantify whether costly sex can be maintained with realistic levels of population subdivision. Simulations in this thesis show that the level of population subdivision (as measured by Fst) needed to maintain costly sex decreases with larger population size; however critical Fst values found are generally high, compared to surveys of geographicallyclose populations. The lowest levels of population subdivision that maintained sex were found if mutation is both advantageous and deleterious, and demes were arranged in a one-dimensional stepping-stone formation. iv) An analytical method is developed to calculate how long it takes an advantageous mutation (such as an invading asexual) to spread through a subdivided population. The flexibility of the methods created means that they can be applied to different types of stepping-stone populations. It is shown how to formulate the fixation time for one-dimensional and two-dimensional structures, with analytical methods showing a good fit to simulation data.