Tail probabilities of sums of heavy-tailed random variables are of a major importance in various branches of Applied Probability, such as Risk Theory, Queueing Theory, Financial Management, and are subject to intense research nowadays. To understand their relevance one just needs to think of insurance companies facing losses due to natural disasters, banks seeking protection against huge losses, failures in expensive and sophisticated systems or loss of valuable information in electronic systems. The main difficulty when dealing with this kind of problems is the unavailability of a closed analytic expression for the distribution function of a sum of random variables. The presence of heavy-tailed random variables complicates the problem even more. The objective of this dissertation is to provide better approximations by means of sharp asymptotic expressions and Monte Carlo estimators. By doing so, we will obtain a deeper insight into how events involving large values of sums of heavy-tailed random variables are likely to occur.