Cellsize

Cell size, also know as resolution, is the size of each individual cell, or square, on an image. These hundreds, thousands, or millions of cells are collectively know as a raster. This is a way of quantifying information in a grid format over an image. For instance, imagery from satellites in many cases have a maximum resolution. This resolution determines how much detail is included as the scale of the image is increased. The resolution of Landsat is ~30m, which means the smallest unit that maps to a single pixel within an image is ~30m x 30m. The latest commercial satellite (GeoEye 1) has a resolution of 0.41m (effectively 0.5 due to US Govt restrictions on civilian imaging). Imagery of the whole Earth are available from many sources, but much of this is of a large cell size. As the cell size decreases the size of the image or raster inversly increases, as well as the computing power needed to use it.

Cell Size in a GIS
The cell size of your data within a GIS could have many impacts on the data and any analysis they are used for. In general an analysis is only as accurate as the largest cellsize used in the analysis. This is because while a cell could be broken up into small cells, it only takes on the attributes of the larger cell, and is in effect still at the larger scale. If an analysis was done with 30m cell sizes, the most accurate the analysis could be is 30m. Analyst must also be conscious of the size and resolution of the raster. The data should be of a small enough resolution to capture the required detail. At the same time it must also not be so small a scale that computing time and power take a significant hit from processing the large dataset.