Resolution independence is the main reason. No matter what the screen density, the images will look sharp. And, there's even the possibility that rendering engines can use their own tricks (such as sub-pixel rendering) to make it look even better (though I'm not sure if any browsers actually do this).
File sizes can be better (though they're occasionally worse), and I'm not doing anything fancy enough to be straining any [CG]PU made in the past decade.
Also, I really like the fact that the image format more closely represents the actual intent of the image creator. "Draw a line from [ax, ay] to [bx, by]" is much more accurate than "Draw a pixel at [ax, ay], then draw a pixel at [ax+1, ay+2], then draw a pixel..."
Few do, but creating SVG on the fly is also much more straight forward than other formats (no need to learn a specific image library). I've seen people do graphing engines which generate an SVG graph. Although this benefit is a lot less beneficial now when we have canvas support in a lot of browsers.
SVG is plain text XML after all. Not binary blobs.
File sizes can be better (though they're occasionally worse), and I'm not doing anything fancy enough to be straining any [CG]PU made in the past decade.
Also, I really like the fact that the image format more closely represents the actual intent of the image creator. "Draw a line from [ax, ay] to [bx, by]" is much more accurate than "Draw a pixel at [ax, ay], then draw a pixel at [ax+1, ay+2], then draw a pixel..."