Because in the natural world data (weights, heights, etc) is distributed normally - by that I mean that the number and magnitude of deviations from the mean are roughly the same on either side of the mean. Thus data forms a bell-shaped curve around the mean. Depending on the data being considered the bell may be tall and thin or short and fat, but it will still be bell-shaped (ie. normally distributed)

Often in our non-natural world data is not normally distributed - it is skewed - for example the number of minutes late a train arrives at a station each morning, the number of children a family has - these are not likely to be normally distributed. Thus other distribution have to be considered.