What you need to know before considering PGD for gender determination:

One, PGD is a complex procedure:
It requires expertise in embryology (making good embryos with high potential to implant), biopsing the embryos to extract one single cell to analyze it genetically, handling of embryos (some doctors are better at handling them than others), and genetics (since only one cell is analyzed, the genetics laboratory has to be highly experienced and have a low frequency of errors). As a patient you should get as much information as possible regarding the error rate of the PGD laboratory the IVF center is using. It should be preferably published in a recent publication, and it should be below 10%, and preferably below 5%. The embryologist doing the procedure should have done it many times and the same goes for the doctor doing the embryo transfer.

Two, biopsy has some negative effect on pregnancy rates:
A very well done study by Goessdens et al. (2008, Human Reproduction, in press) shows that biopsing two cells results in lower chance of achieving a pregnancy than biopsing one cell. From that one can deduct that biopsying zero cells is probably better than biopsing one. If you have decided to undergo PGD for gender selection you are accepting a small decrease in your chances of conceiving.

Three, the biopsy effect can be compensated by the genetic screening of 9 or more chromosomes:
In women 35 and older, 50-80% of the embryos are chromosomally abnormal (Munne et al. 2007, Reproductive Biomed Online 14:628-634). The method used by most IVF to select the best embryos for transfer is embryo morphology and/or blastocyst culture, but that may improve the odds of transferring a normal embryo only to 60%, the other 40% being abnormal.
Although some studies have shown that PGS (PGD analyzing at least 9 chromosomes, instead as 2-5 for gender selection) does not compensate for the damage of the biopsy, in experienced hands PGD testing for 9 or more chromosomes does seem to improve the odds of conception. Why? Because abnormal embryos (40-80% of all embryos) either do not implant or miscarry. Why go through this expensive, stressful, and frustrating procedure to end up losing your valuable pregnancy? Without PGS, women 35-40 years will lose 18-26% of their IVF pregnancies, and women 41-42 years will lose 40% of them. After PGS the loss is significantly reduced, to 14% in women 35-40 and 22% in women 41-42 years old (Munne et al. 2006, Reproductive biomedicine online 12:234-253). In addition, if the biopsy is done well, and the PGD lab has a low error rate in the genetic test, PGD can improve implantation rates by selecting normal embryos with higher odds to implant (Munne et al. 1999, 2003, Gianaroli et al. 1999, Colls et al. 2007 Fertility and Sterility 88:53-61).
However, while the 9-chromosome test can detect 61% of aneuploidías (abnormalities), the 12 can detect about 70% and array CGH can detect 100% of aneuploidías, the 5 test can only detect 14%!!!, and a X and Y test less than 10%.
Indeed, the miscarriage rate in PGD for gender selection in women younger than 40, was 0% when using 9 or 12 probes and 15% when using 3 or 5 probes.

Four, if you decided to go for gender selection it pays to do PGS for 9 to 12 chromosomes or array CGH:
So you have decided you want to do gender selection. It pays to analyze not only those chromosomes involved in gender determination, but also those that can make an embryo not to implant or abort, because:

1. If there is damage done by the biopsy, the damage is already done before the genetic test. Thus, I recommend you to chose a test that can give you more information, and probably better odds to conceiving than a simple XY test.
2. By analyzing 9 or more chromosomes you may have a significantly higher chance of conceiving and keeping the pregnancy to term.
3. You will have 5-fold reduced chance of conceiving a pregnancy that could reach term but have Down syndrome or other devastating syndromes, such as trisomy 13 or 18, which usually result in baby death within a year. Actually the risk of conceiving an abnormal pregnancy after PGD for a women 35 or older is reduced to the risk of a much younger women, for whom prenatal diagnosis (CVS or amniocentesis) is usually not recommended. The cost of doing gender selection (with 2-3 chromosomes) compared to 9 chromosome PGS is thus similar if you consider that prenatal diagnosis may not be needed.
4. Granted, the more chromosomes are analyzed, the higher the error rate of analyzing a single cell. However, all the improvement in implantation, pregnancy loss prevention, and abnormal offspring prevention has been achieved with 9 or more chromosomes tested, so that risk is already factored in. Nevertheless, you should seek PGD laboratories with published error rates that are way below 10% (example, Colls et al. 2007, Fertility and Sterility 88:53-61, error rate 4.7%).
5. Because on average only 30% of embryos analyzed for 9 to all chromosomes will be classified as normal, the number of embryos of the unwanted gender classified as abnormal will be lower, reducing the ethical burden of the procedure (see below point).

In conclusion, for 500-700$ difference that cost a 9 probe test compared with a 3 or 5 chromosome test (and provided the test is performing the test in the right IVF lab and the right PGD lab), by analyzing 9 or more chromosomes you get an increased chance to carrying the pregnancy to term, of the embryos implanting, a reduced risk of conceiving a Down syndrome or several other syndrome babies, and you may not need afterwards to pay for prenatal diagnosis.

Five, you should consider the ethics of gender selection:
Gender selection is considered by some to be unethical (and unlawful in Europe) because it either can lead to gender discrimination or destruction of normal embryos of the unwanted gender. Reprogenetics has studied the use of PGD for gender diagnosis (Colls et al, 2009, Reproductive Biomedicine Online, in press) and has reached the conclusion that at least in the US there is no risk of gender discrimination because the technique is mostly used for family balancing. However, discarding perfectly normal embryos, Reprogenetics strongly encourage couples seeking this procedure to select these embryos for chromosome abnormalities with 9 or more chromosomes tested, and to freeze or donate the normal embryos of the unwanted gender.