Digital watermarking is a technique that aims to embed a piece of information permanently into some digital media, which may be used at a later stage to prove owner authentication and attempt to provide protection to documents. The most common watermark types used to date are pseudorandom number sequences which possess a white spectrum. Chaotic watermark sequences have been receiving increasing interest recently and have been shown to be an alternative to the pseudorandom watermark types. In this paper the performance of pseudorandom watermarks and chaotic watermarks in the presence of common watermark attacks is performed. The chaotic watermarks are generated from the iteration of the skew tent map, the Bernoulli map and the logistic map. The analysis focuses on the watermarked images after they have been subjected to common image distortion attacks. The capacities of each of these images are also calculated. It is shown that signals generated from lowpass chaotic signals have superior performance over the other signal types analysed for the attacks studied.