First make sure you're really good with multiplication and division *in your head*. It helps a lot when doing steps of the next parts.
Factoring numbers *in your head* is an example of something very helpful.
Then make sure you understand algebra. Parentheses, simplification, extraction of terms.
Move on to algebraic equations -- solve for one variable in linear and quadratic equations. Solve for two variables in a two-equation system.
Then do trigonometry -- sine, cosine, tangent, and the inverse of those functions. Understanding trig is super important, but once you're into linear algebra and do 3D math, it's actually not *used* much, just a good grounding to understand.
A traditional curriculum would then lead you through limits, derivatives, and integrals -- classical calculus of one variable. On the one hand, that's useful for later areas. On the other hand, it's not useful for much *real*.
I'd rather just do linear algebra at that point. Vectors, matrices, quaternions as they pertain to 3D geometry.
Another useful set of math is the intersection with computer science you get in computational geometry. Anything from graph theory, set geometry, to spatial indexing. This requires the linear algebra to become efficient.
If you have calculus you can move on to differential equations, and multi-variable calculus. Diffs are useful for motion systems and other continuously evolving systems -- analog electric circuits, mechanical motion, etc.
Finally, the red-headed step children of applied mathematics: numerical methods, and statistics. Both are actually important, and neither gets much love from the mainstream.
After that, you're approximately OK for a BS in math. From there, you can go into applications (which typically intersect with computer science, and/or perhaps EE or Mech) or specialize into theoretical math.
Good sites:
http://mathworld.wolfram.com/http://www.khanacademy.org/