Rules. Who needs them? Well, if you never share your work with anyone else and your models are less then 10 features, then you can probably ignore this. For everyone else, this might be useful. Pro/Engineer (and other parametric CAD programs) can create hugely complex structures that will definetely bite you back if you don't carefully manage them well.

I've broken down these rules into general strategies and more specific tactical rules.

1. Simplify, simplify, simplify.
   Since Pro/E creates complex structures, you should be going out of your way to keep everything else simple: your database organization, your assembly structure, and part design. This rule should especially be considered when determining whether to use Layouts, Copy Geom, Inheritance, Family Tables, etc.
2. Develop Good Habits
   This takes discipline. What may help is if you put in place a system of continual training. This doesn’t necessarily mean paid classroom sessions with a tutor – although that would be ideal. You may simply want to put aside 2 or 3 hours every week for studying a free tutorial or suggested methodology.
3. Focus on Quality
   A lot of the time, when under pressure, we tend to focus on simply finishing, rather than doing things right. This is short-term thinking. Instead, focus on the lifecycle of the model. You may think you're just building a packaging study, but what happens if it evolves into the data for a prototype, and then a Master Model, and finally ends up cutting steel for a mould.
4. Planning
   Before doing any modeling, spend an appropriate amount of time planning your model build. There are always 10 ways of building a model. Consider the pros and cons of each and, if possible, discuss strategies with a colleague. With a little bit of experience you can build a model on paper and figure out what kind of issues you might run into. Only when you have a decent plan should you start modeling.
5. Learn to Think like a Designer
   This applies only to consumer design and any objects that use an Industrial Designer. Unless you become aware of what makes designers tick, you are always going to be frustrated. You will have to rebuild surfaces over and over until you get it right! When you think you might be finished, take a look at the surfaces under a microscope. Use surface tools like Reflection Maps, Gaussian, and Porcupines to evaluate quality.

1. Simplify, simplify, simplify.
   Hmm. I've seen this before somewhere! Well, simplification is equally important at the feature and model level. For example, why create one complex curve sketch, when you can create two or three simple ones? While we're in the sketcher, geometric features (arcs, ellipses) are more manageable than splines. Use them if you can. Having said that, when you need a spline (and who doesn't), there is no substitute.
2. Minimize Feature Count.
   I put this one here just to contrast with the previous rule. In general, small feature count models will be cleaner and better organized. They will probably not contain redundant, unused features or low-quality fill and cut geometry. Having said that, don't reduce features for the sake of it. The point made in Rule 1 still applies.
3. Build Models Twice
   Your boss may not like the sound of this one, but it will probably save your company valuable time in the long run. I've found that it is convenient to create a quick and dirty model first. From this you can discover many potential issues without having to consider quality issues. Warning! This method can be dangerous if you end up having to develop your initial model for production. Don't spend more than a few days on it. By the way, this does not conflict with Planning (Rule 4 above), since the first model build can be considered part of the Planning process.
4. Learn from History
   Your model tree isn't just there to enable feature redefinition. It also provides valuable clues regarding model construction. Make your model tree smart. Use feature naming and intelligent grouping to provide logical history. For example, surface Merges should be placed immediately after the surface they consume (where possible).
5. Build in Flexibility
   This is probably the most frequently stated golden rule for Pro/E. There is no point in using a parametric modeler if it is not easy to modify. There are a few rules for flexiblity such as overbuilding and avoiding complex sketches, but mostly you just need common sense. Think about whether to tie each feature to another feature or whether it should be simply dimensioned to a datum plane.
6. Build Large Features First
   Create large surfaces first – small ones later. This is a general rule that mostly holds true. An analogy would be building construction. Stud walls, roof trusses, sheathing, drywall go up before smaller details such as door hardware, and plumbing.
7. The Fewer Parents the Better
   This is not a hard and fast rule, but one worth knowing about. The more parents a feature has, the more "dirty" its geometry is likely to be. This partly due to tolerance issues and partly due to the compounding complexity. For example, if you create a datum curve, then project this onto a complex surface, then use this second curve as a reference within a section, what you finally end up with may not be as perfect as you might want. Take a good look at the porcupine curves to verify this.
8. Layers are your Best Friends
   Layers are there for convenience. If your company has an excessively elaborate or confusing layering scheme, ask them to simplify it. There are many times when users feel that they are out of their depth with Surfacing simply because of the confusion on the screen. Your layers should enable you to clean up the screen and isolate the area you're working on with a few clicks of the mouse.
9. Build Draught into Features
   The Pro/E draught feature is useful for simple geometric features, but not much use for anything else. If you have a moulded part, draught should always be implemented as part of the feature creation process. This can normally be done quite readily whether you're creating an extrusion, a revolve or a sweep.
10. Create Inner Quilts Late
    This only applies if you have a thinwall part. By creating an offset for the inner wall as late as possible, you will have a surface that reflects as much of the external geometry as possible. There comes a time in the models history when it is no longer possible to create the desired offset. Your offset should be immediately before that point.