I think that the main advice I'd give is this: keep it simple. Once you've mastered the basics (lathes, extrusions, pathed extrustions, and really simple lofting)... everything past that gets pretty tricky, and it will take time to get any good. Start off with very simple projects... that will show you how to do some of the most difficult things. And don't neglect to do a lot of reading about the theory behind the applications- knowing what a surface normal is and how it works, for example, is very useful, even if it seems hopelessly esoteric at first.
If you're wanting to learn organic modeling, I suggest trying to model an apple. It's cheap to go get a scale model you can stare at, and you can eat a snack when you're done
![](smilies/icon_smile_wink.gif)
Apples are actually surprisingly difficult to model, because of their complex curvature, where the bumps on the bottom are, and will teach you a great deal about the technical challenges of organic modeling. If you learn how to do an apple, then modeling something really complicated, like a human head, becomes much more possible. But don't expect to do lifelike heads right off the bat- you'll just get frustrated.
If you're wanting to model mechanical things... I suggest trying to model a ship (y'know, the ocean-going kind) as your first project. Ships have lots of conveniently flat surfaces and simple shapes piled on top of each other, and you can have a lot of fun learning how to make greebles (the small details that make a thing believable). Work from a photo, or better yet a 3-view or 4-view drawing (such things can be found all over the Internet). Try for maximum accuracy- building fantasy ships is easy... building one that is like a real thing is not!
If you have a specific goal in mind... then you're going to need to practice for awhile before you can reach it. I often see people, when they're just starting out, build things that look really cool... but have 100,000 polygons, when they could've modeled them efficiently with 20,000. That's the difference between a neophyte and a master- understanding enough about the limitations of rendering engines (whether real-time, for games or other multimedia, or raytracers) to do a good job, fast, without wasting any polygons. Many people take over a year to get really good with 3D software... so don't be surprised if your first efforts make you wince later... it's happened to all of us
Texture-mapping (skinning) is a whole 'nother subset of skills, and requires some specialized tools and some practice to get any good. But if you can't build geometry, then there's no point being able to skin- understanding one is key to understanding the other. But a good skin can make even the crappiest models look surprisingly good.