A dimension is the intended size of a part, while a tolerance is how much that size is allowed to vary. Every manufactured component has both: the dimension tells you what the target measurement is, and the tolerance tells you how close to that target the finished part needs to be. Together, they define what counts as an acceptable part and what gets rejected.
Dimensions: The Target Number
A dimension is simply the numerical value that defines the size or geometric characteristic of a feature. When an engineer draws a shaft that needs to be 25 mm in diameter, that 25 mm is the dimension. Dimensions describe lengths, widths, heights, hole diameters, angles, and the distances between features. They represent the ideal version of the part, the one that would exist if manufacturing were perfect.
In practice, no manufacturing process produces a part at exactly its nominal dimension every single time. A CNC machine, injection mold, or lathe will always introduce tiny variations from one part to the next. That’s where tolerances come in.
Tolerances: The Acceptable Range
A tolerance is the total amount a dimension is allowed to vary. It’s defined as the difference between the upper (maximum) and lower (minimum) limits of a dimension. If a 25 mm shaft has a tolerance of ±0.05 mm, any shaft measuring between 24.95 mm and 25.05 mm passes inspection. Anything outside that 0.10 mm window gets scrapped or reworked.
There are two main ways tolerances are expressed on engineering drawings:
- Bilateral tolerance allows variation in both directions from the nominal value, typically written as ±0.05. The part can be slightly larger or slightly smaller. This is the more common and flexible approach, well suited to brackets, covers, and non-critical features.
- Unilateral tolerance restricts variation to one direction only. A dimension might read +0.00/−0.10, meaning the part can be up to 0.1 mm smaller than nominal but not any larger. This is common for press-fit shafts or pins where exceeding a specific diameter would prevent assembly.
Bilateral tolerances can also be unequally distributed. A part might allow +0.03/−0.07, giving more room for variation in one direction than the other. This is useful when the function of the part is more sensitive to growth in one direction.
Why Tolerances Exist
Tolerances exist because perfection is impossible and, more importantly, unnecessary. Most parts don’t need to be exact to function correctly. A bracket holding a cable in place might work fine with a full millimeter of variation, while a fuel injector needle might need tolerances measured in thousandths of a millimeter. The tolerance tells everyone involved, from the machinist to the quality inspector, how much precision actually matters for that specific feature.
Choosing the right tolerance is a direct cost decision. Tighter tolerances require slower machining speeds, stronger fixturing, climate-controlled environments to manage thermal drift, more advanced inspection equipment, and higher operator skill. Every extra micron of precision increases scrap and rework risk. One manufacturer found that by relaxing a tolerance on non-critical mounting holes from ±0.01 mm to ±0.02 mm, they cut inspection time by 35% and significantly reduced machining costs. The parts worked identically in the final assembly.
How Tolerances Determine Fit
When two parts need to connect, like a shaft sliding into a hole, the tolerances on both parts determine the type of fit. There are three categories:
- Clearance fit: There is always some gap between the shaft and hole, even when the shaft is at its largest allowed size and the hole is at its smallest. The parts slide together freely.
- Interference fit: There is always some overlap. The shaft is slightly larger than the hole, so the parts must be pressed or heated into place. This creates a tight, permanent connection.
- Transition fit: Depending on where each part falls within its tolerance range, the result could be a slight gap or a slight press. You won’t know until the parts are made.
International standards (ANSI/ASME B4.1 and ISO 286) define specific classes of fits within each category, giving engineers a standardized way to specify how tightly two parts should mate.
Tolerance Stack-Up in Assemblies
When a product has multiple parts that fit together, each part’s tolerances add up. This is called tolerance stack-up, and it’s one of the most common sources of assembly problems. If you’ve ever put together flat-pack furniture and found the last screw won’t align, you’ve experienced it firsthand. A tiny imperfection in one panel combined with a slightly misplaced hole in another can throw the entire piece off.
Engineers use stack-up analysis to predict this before production starts. They add up all the individual tolerances across every part in an assembly to calculate the worst-case total variation. A small miscalculation can lead to parts that don’t align, buttons that stick, or enclosures that won’t close. Critical interfaces like a snap-fit lid need tight control, while internal support brackets might tolerate much more variation. The goal is to assign tight tolerances only where they matter and keep everything else loose enough to be cost-effective.
Geometric Dimensioning and Tolerancing
Basic coordinate dimensions (length, width, distance from an edge) work well for simple parts, but they have a significant limitation: they don’t tell the inspector how to set up the part for measurement. Two inspectors measuring the same part with coordinate dimensions can get different values simply because they oriented the part differently.
Geometric Dimensioning and Tolerancing, or GD&T, solves this by introducing datums, which are reference planes, axes, or points that establish exactly how the part should be positioned during inspection. This eliminates ambiguity. GD&T is governed by the ASME Y14.5 standard, which defines the symbols, rules, and practices for stating and interpreting tolerances on engineering drawings and digital models.
GD&T also produces more practical tolerance zones. With coordinate dimensions, the tolerance zone for a hole’s position is a square. But since holes and the bolts that go through them are round, a square zone is unnecessarily restrictive in the corners. GD&T uses a circular tolerance zone instead, which is 56% larger in area than the equivalent square zone. That means more manufactured parts fall within spec without any loss in function, reducing scrap and lowering production costs.
Measuring Tools Match the Tolerance
The precision of your measuring tool needs to match the tolerance you’re checking. A standard caliper reads to 0.01 mm (or 0.001 inches), with a typical accuracy of ±0.02 mm. That’s fine for tolerances in the tenths of a millimeter. For tighter tolerances, a micrometer reads to 0.001 mm (one thousandth of a millimeter), which is ten times more precise than a caliper.
As a general rule, your measurement tool should be at least ten times more precise than the tolerance you’re verifying. If a part has a tolerance of ±0.1 mm, a caliper works. If the tolerance is ±0.01 mm, you need a micrometer or something even more specialized. Using the wrong tool doesn’t just risk bad measurements; it can make it impossible to tell whether a part is actually in spec.