Janice Fraser at Adaptive Path has a list of tips for setting priorities:

1. Make a "Big List of Things to Do"
2. Organize the list according to Dependencies (can't do X until Y) and Baseline items (absolutely, positively must be done)
3. Have the appropriate co-workers score each item according to:
• Technical feasibility
• Creative feasibility
• Importance to the user
• Importance to the business
4. Graph the overall scores

A task that shows up with both high feasibility and high importance scores should be done first. Something that gets low feasibility and low importance scores probably shouldn't be done at all. The tricky tasks will be those that fall in the category of: high importance/low feasibility meaning that they're important, but difficult.