A convenience sample is a sample of study subjects taken from a group which is conveniently accessible to a researcher. The advantage of a convenience sample is that it is easy to access, requiring little effort on the part of the researcher. It is not an accurate representation of the population, skewing the results quite radically and rendering any conclusive data hard to make convincing and from which to draw generalisations, regarded as a form of sampling bias, meaning that the results from a study conducted with such a sample cannot be generalised to the population as a whole. It is to be used when unable to access a wider population due to constraints in budget or time. As the name suggests, one should not worry about taking random samples of the population; just use people who are available, whom you know or meet, and use as many people as possible to ensure results from a single test is not just a coincidence. There are very few rules governing how the sample should be collected, and relative to other sampling techniques the cost and time is small; indeed, other techniques may require more formal access to lists of people from whom to select for a survey. Convenience sampling is often used in pilot studies to allow the people conducting the study to obtain basic data and trends regarding a study without such unwieldy complications and can also document a particular quality of a substance or phenomenon occuring within a given sample. To interpret the findings from a convenience sample properly, you have to characterise (usually in a qualitative sense) how your sample would differ from an ideal sample that was randomly selected. In particular, attention should be paid to who might be left out of a convenience sample or who might be underrepresented in your sample.
Convenience sampling is often a preferred option to other methods of sampling because it allows an experimenter to pilot-test an experiment with minimal resources and time. It is also relatively inexpensive and allows the researcher to get a gross estimate of the results.