Here's a citation:
Garland CF, French CB, Baggerly LL, Heaney RP. Vitamin D supplement doses and serum 25-hydroxyvitamin D in the range associated with cancer prevention. Anticancer Res. February 2011;31:607-11.
Cancer prevention is correlated most closely with serum levels of 40 ng/mL or above (as is alleviation of depression), and to reach this level in 97.5% of the population, 9,600 IU/day was necessary. This is almost double the current UL. Of course, current recommendations for daily dosage is based off 20 ng/mL being 'sufficient', while most experts in the field now believe that 30 ng/mL should be the baseline for 'sufficient' and that most positive effects will be found at serum levels of 40 - 50 ng/mL.
Vitamin D toxicity is rare, and only occurs when serum 25(OH)D levels exceed 150 ng/mL. It has never been demonstrated at doses of less than 20,000 IU/d and generally requires greater than 40,000 IU/d. Most incidents have been due to accidental ingestion, such as from a milk supply that was accidentally fortified with vast amounts of D3.
In the end, supplementation needs to be based off serum 25(OH)D levels, which can be measured by a doctor. You may need more or less to reach 'ideal' levels, and it's impossible to say exactly how much without testing. The test is cheap and hopefully will become a standard part of a routine examination, considering that vitamin D affects at least 35 different systems in the body. Without the test, 2,000 IU/d will keep you under the UL (even though it should really be changed to reflect the science behind the toxicity), and will likely keep your levels above 30 ng/mL. Remember that most dairy products are fortified with D3 which should be considered a part of the total.