Discussion of a Norwegian study looking at 50,000 people who didn't have pre-existing heart disease for ten years. http://drmalcolmkendrick.files.wordpress.com/2012/09/mortality-and-cholesterol1.png?w=600&h=309

As you can see, for women the story is very straightforward indeed. The higher the cholesterol level, the lower the risk of overall mortality. With regard to heart disease alone, the highest risk is at the lowest cholesterol level. For men there is more of a U shaped curve, but overall mortality is highest at the lowest cholesterol level.

Here's the study.

Here's the actual conclusion from the study, which dhoe pointed out in comments is considerably milder than the quote above:

Based on epidemiological analysis of updated and comprehensive population data, we found that the underlying assumptions regarding cholesterol in clinical guidelines for CVD prevention might be flawed: cholesterol emerged as an overestimated risk factor in our study, indicating that guideline information might be misleading, particularly for women with ‘moderately elevated’ cholesterol levels in the range of 5–7 mmol L-1. Our findings are in good accord with some previous studies. A potential explanation of the lack of accord between clinical guidelines and recent population data, including ours, is time trend changes for CVD/IHD and underlying causal (risk) factors.

‘Know your numbers’ (a concept pertaining to medical risk factor levels, including cholesterol) is currently considered part of responsible citizenship, as well as an essential element of preventive medical care. Many individuals who could otherwise call themselves healthy struggle conscientiously to push their cholesterol under the presumed ‘danger’ limit (i.e. the recommended cut-off point of 5 mmol L-1), coached by health personnel, personal trainers and caring family members. Massive commercial interests are linked to drugs and other remedies marketed for this purpose. It is therefore of immediate and wide interest to find out whether our results are generalizable to other populations.

However, the chart (the png link above-- I don't know how to make the image appear) shows that the all cause mortality for women was lower if their cholesterol results were higher.

A different big study which also found that low cholesterol was dangerous, but high cholesterol was also dangerous in terms of heart attacks, though mostly for men under fifty, and (I think) not so much for women.

A comment explains that the usual test for cholesterol isn't actually for cholesterol, it's for the lipoproteins which keep all sorts of fat molecules from forming large blobs in a watery environment.

This sort of thing appeals to a number of my prejudices, so I'm hoping to get some more meticulous angles on it from LW.

Post edited to add discussion of the conclusion of the Norwegian study.