This site which is very pro-Atkins, anti-Ornish has almost too much information, but it's a nice explanation of Syndrome X, Setpoint theory, and understanding research papers.
http://www.omen.com/adipos.html
A sample of the information:
Ornish and Atkins Compared
We need large scale randomized studies comparing low fat and low carbohydrate diets. Until such studies is published, we must compare results reported by Ornish and Atkins themselves. The Ornish figures are the average of the Ornish Experimental group (n=22) carefully selected from hundreds of applicants. The Atkins data (n=1) is from page 150 of his 1992 book. Neither of these samples is necessarily representative of the overweight population. However, the starting age, weight, and body mass index of Atkins' sample resembles those of Ornish's experimental group much more closely than Ornish's own control group, lending credence to the comparison.
The changes in metabolic risk factors agree with those reported in the METABOLISM and DIABETES CARE studies discussed above, suggesting the differences between low-fat and low-carbohydrate diets reported in those papers are applicable to a wider population.
Ornish and Atkins Diets Compared
INITIAL CONDITIONS ORNISH (low fat) ATKINS (low carb)
Age 56 55
Starting weight 201 195
Body Mass Index 28.4 28.1
DIET RESULTS ORNISH (low fat) ATKINS (low carb)
Cholesterol change -24% -13%
HDL (GOOD) Cholesterol - 3% +60%
Triglycerides (BAD) +75% -82%
Weight -12% -19%
Further confirmation of Atkins' data comes from Dr. Ronald Krauss, chairman of the American Heart Association's Nutrition Committee. In some men with normal-sized LDL cholesterol particles, a very low-fat diet can cause changes in the cholesterol profile that indicate an increased heart disease risk. In one study, 36 out of 87 men with normal-sized LDL particles switched to the small-particle abnormalities when their dietary fat was lowered from 46 percent of calories from fat to 24 percent. One sign of the switch was a rise in their ratio of total cholesterol to "good" HDL cholesterol, which implies a higher heart disease risk. (API 7/16/96)