A qPCR instrument hands you a curve before it hands you a number. The number, the Cq, is derived from the curve, and it is only as trustworthy as the curve it came from. Most people learn to read the Cq off a table and never look at the shape that produced it. That is a mistake, because almost everything that can go wrong in a run is visible in the curve first, and a Cq read from a bad curve is a confident answer to the wrong question. This article is about reading the curve itself: what its three phases mean, where the threshold belongs, and what a Cq actually measures.
The shape of a healthy curve
Plot fluorescence against cycle number for a reaction that worked, and you get a sigmoid, an S laid on its side. It has three parts, and each part is telling you something about what the polymerase is doing.
In the **baseline** phase, product is doubling every cycle, but there is so little of it that the fluorescence signal is lost in the background noise of the instrument and the reaction chemistry. The curve looks flat here not because nothing is happening but because a few thousand copies cannot be seen against the floor. This region, typically the first ten to fifteen cycles, is what the software uses to learn what "zero" looks like for each well.
The **exponential** phase is the part that carries the quantitative information. Once product accumulates enough to clear the noise, the curve turns sharply upward, and for a stretch of cycles it rises with a steep, near-constant slope. Here the reaction is still reagent-rich and efficient, so each cycle roughly doubles the product, and the cycle at which a given well reaches a fixed amount of product depends directly on how much template it started with. This is the only phase where the numbers mean anything.
The **plateau** phase is where the curve flattens at the top. Primers and nucleotides deplete, the polymerase competes with product reannealing, and amplification grinds to a halt. The height of the plateau is not proportional to starting template and should never be used for quantification. Two wells that started decades apart in concentration can plateau at nearly the same height. The plateau tells you the reaction finished, and little else.
Baseline and threshold: how the instrument finds a number
To turn that shape into a Cq, the software does two things. First it subtracts the baseline: it fits the flat early-cycle region for each well and sets it to zero, which flattens out the well-to-well differences in background fluorescence so every curve starts from the same floor. Get the baseline window wrong, and you distort every curve built on top of it, so this step matters more than it looks.
Then it places a **threshold**: a horizontal line at a chosen fluorescence level. The Cq, the quantification cycle, is simply the cycle number where a well's curve crosses that line. A common default sets the threshold about ten standard deviations above the baseline noise, and the baseline window is set to end a couple of cycles before the earliest-crossing sample so that no real signal leaks into the region the software treats as flat.
The one rule that governs threshold placement is that the line has to sit inside the exponential phase of every curve you intend to compare.
Set it too low, down in the baseline noise, and the crossing point is decided by random fluctuation rather than by amplification, so your Cq jitters from replicate to replicate. Set it too high, up in the plateau, and the curves have already bunched together, so real differences in starting amount collapse and the crossings land almost on top of each other. Inside the exponential band, where the curves are steep and well separated, a small change in threshold height barely moves the Cq, which is exactly the robustness you want. Modern instruments place the baseline and threshold automatically and usually get it right, but "usually" is why you still look.
What a Cq actually measures
Here is the part that trips people up: a Cq is a relative measurement, not an absolute one. It counts how many cycles a reaction needed to reach a fixed amount of product. More starting template means fewer cycles to get there, so a lower Cq means more target. Because the growth is exponential, the relationship is logarithmic: each tenfold change in starting template shifts the Cq by a fixed number of cycles.
For a reaction that doubles perfectly every cycle, that spacing is about 3.3 cycles per tenfold, because two raised to the 3.32 is ten. That single fact is why a dilution series in qPCR marches across the plot in even steps, and why those steps are the basis of every standard curve. It is also why a raw Cq, on its own, is close to meaningless. A Cq of 22 is not a quantity; it only becomes one once you know the efficiency of the reaction and have something to compare it against, a standard curve or a reference sample. Two assays with different efficiencies can report the same Cq from genuinely different amounts of template.
The instrument reports a Cq for every well, but only the curve tells you whether that number deserves to be trusted.
Reading the curve is therefore not an optional nicety on top of reading the Cq. It is the check that tells you the Cq is real: a clean flat baseline, a steep and well-separated exponential rise, a definite plateau, and a threshold sitting squarely in the exponential phase. When those four things are present, the number means what you think it means. When any of them is missing, the shape will show it long before the Cq does, which is the subject of the companion piece on curves that look wrong.