The short answer
Where fasting is required, the usual instruction is eight to twelve hours without food. What has changed is the list it applies to. Fasting glucose still needs it. Most of the rest does not, and that includes the lipid panel under current European guidance.
The reason is mechanical. Digestion pushes glucose and triglycerides into the blood within minutes and keeps them raised for hours. Measuring fasting glucose after a meal measures your breakfast, not your metabolism. Nothing in a red cell count moves that way, which is why a complete blood count, a thyroid panel or a liver panel can be drawn at any time of day.
One rule beats every table on the internet, this one included: follow the instruction printed on your prescription, or the one your laboratory gave you. If it says fasting and you have eaten, say so at reception rather than letting the sample go through. A fasting glucose read on a fed sample is not a slightly imperfect result — it is an uninterpretable one.
What fasting actually means
It means no food and no drink other than plain water, for the stated number of hours, right up to the draw. It does not mean simply having skipped breakfast, and it does not mean the last meal was merely taken the evening before.
That distinction is not obvious to most people. When researchers questioned 150 outpatients arriving for blood tests, only 39% could give the correct definition of fasting, and 46% believed the exact interval since the last meal did not matter. More than half had received no instructions at all, and only 60% arrived properly prepared.
So if nobody explained it to you, you are in the majority. The practical version: finish eating in the evening, drink water freely overnight and in the morning, and take nothing else until the needle is in.
Fasting by test type
Check against your own prescription: protocols still differ between laboratories, and some have not yet aligned with the current guidance on lipids.
| Test | Fasting | What changes if you eat |
|---|---|---|
| Fasting glucose | 8–12 h | Rises within minutes of a meal. The test loses its meaning. |
| Lipid panel | Not routinely | European guidance recommends non-fasting testing. Triglycerides rise by about 26 mg/dL; LDL and HDL barely move. |
| Complete blood count | None | Cell counts are not affected by a meal. |
| Liver panel | None, usually | Stable, unless a very fatty meal was just eaten. |
| Kidney panel | None | Creatinine barely moves; hydration matters more. |
| Thyroid panel | None | TSH follows a daily rhythm, not your meals. Prefer the morning. |
| Iron and ferritin | Morning, fasting preferred | Iron varies over the day; a supplement taken that morning distorts it. |
| HbA1c | None | Reflects three months. A meal is invisible to it. |
The lipid panel: the rule most people are still given is out of date
For decades a cholesterol test meant an overnight fast. Since 2016, the European Atherosclerosis Society and the European Federation of Clinical Chemistry and Laboratory Medicine have jointly recommended the opposite: measure lipids without fasting, as the default.
The reasoning is that the differences are small and the fasting state is the unusual one. After an ordinary meal the maximal average changes are about +26 mg/dL for triglycerides, −8 mg/dL for total cholesterol and −8 mg/dL for LDL cholesterol. HDL cholesterol, apolipoprotein B and lipoprotein(a) are not affected at all. Non-fasting and fasting values predict cardiovascular disease equally well, and most of the day is spent in the non-fasting state anyway.
Fasting is still considered in one situation: when non-fasting triglycerides come back above 5 mmol/L, or 440 mg/dL. Denmark, the United Kingdom, Europe, Canada, Brazil and the United States have all endorsed non-fasting lipid profiles in some form. If your laboratory still asks you to fast for cholesterol, that is its protocol, and you should follow it — but the requirement is no longer the international default.
Water yes, coffee no
Water is not merely allowed, it helps: a well-hydrated vein is easier to draw from, and a sample taken in a dehydrated state can slightly concentrate several values. Drink normally, including on the morning of the draw.
Coffee is a different matter, black and unsweetened included. It is not metabolically inert. In a crossover trial, healthy volunteers given caffeinated coffee before a meal showed a glucose response 147% to 216% larger than the same people given decaffeinated coffee, with insulin sensitivity down by up to 40%. In another controlled trial, coffee on its own raised the insulin response compared with water.
That is enough to disturb precisely the test fasting exists to protect, which is why laboratories keep coffee out of the fast. Tea, juice, milk, soft drinks and chewing gum fall under the same rule. So does smoking: nicotine has a metabolic effect of its own, so leave it aside in the hour before the draw.
Medicines: do not stop anything on your own
Take your usual treatment unless the prescribing physician has told you otherwise, in writing. Interrupting an anticoagulant, an antihypertensive or a thyroid replacement to make a number look better carries a real risk, while the number itself only carried an inconvenience.
Two practical points. Levothyroxine is usually taken after the draw so it does not distort the reading. Iron and vitamin supplements are worth pausing for a few days before iron studies — ask your laboratory how long. For everything else: carry on, and tell the laboratory what you take.