Fixed vs Variable Cost Calculator (High-Low Method)

Background.

Deciding whether a particular cost is fixed or variable is a judgement, not a calculation. OpenStax's cost behaviour chapter says so directly: a single cost such as rent may be classified by one company as a fixed cost, by another as a committed cost, and by a third as a period cost, depending on the decision it is being used for. No calculator can look at your utility bill and tell you which bucket it belongs in.

What can be calculated is the version of the question that has an arithmetic answer. Most real costs are not purely one thing or the other — they are mixed, with a standing charge that does not move and a usage component that does. Your electricity bill, your warehouse costs, your customer-support payroll, your maintenance spend: all of them have a floor and a slope. The high-low method separates the two from two observations of the same cost at different activity levels, and it is the standard technique taught for exactly this purpose.

The mechanics are simple enough to check on paper. Take the period with the highest activity and the period with the lowest, subtract the low cost from the high cost, and divide by the difference in activity. That slope is your variable cost per unit. Multiply it by the activity in either period, subtract from that period's cost, and what is left is the fixed component. Together they give the cost equation Y = a + bx, where a is fixed cost, b is variable cost per unit and x is activity. This page returns a and b, uses them to predict total cost at any volume you choose, and reports what proportion of that predicted cost is fixed versus variable.

The method's weakness is stated plainly by its own sources and you should treat it as a hard limit rather than a footnote. It uses two data points and ignores everything in between. If your highest-activity month was also the month a machine broke, or your lowest was the month you settled an annual insurance premium, that single anomaly sets the slope for the entire estimate. Both authorities behind this page say the same thing: the estimate is based on only the highest and lowest levels of activity, and those points may not represent the data set as a whole. Before trusting the output, look at the two rows you fed it and ask whether either was unusual for a reason that has nothing to do with volume. If one was, use the next-highest or next-lowest ordinary period instead, or move to regression across all your months.

Two results deserve a specific warning. A negative fixed cost is arithmetically valid and economically nonsense: it means cost rose faster than proportionally between your two points, so the fitted line crosses the axis below zero. The calculator returns it rather than hiding it, because it is the clearest possible signal that a straight line does not describe this cost. And a prediction far outside the activity range you observed is not really a prediction at all — cost behaviour is only linear inside the relevant range, and beyond it a second shift, another lease or a volume discount will move the line in ways two data points cannot see.

Use this page to get a defensible first split of a mixed cost, then sanity-check it against what you know about the business. It is a starting estimate, not an audit.

What is fixed vs variable cost calculator?

The high-low method estimates the fixed and variable components of a mixed cost from two observations: the periods with the highest and lowest activity. OpenStax's Principles of Accounting, Volume 2 states the variable rate as "cost at the high activity level minus cost at the low activity level divided by highest activity level minus lowest activity level", and derives the fixed component as "fixed cost = total cost − variable cost" evaluated at either observed point. The result is expressed as the cost equation "Y equals a plus bx", where "a is the fixed cost" and "b is the variable cost per unit, and x is the level of activity". A mixed cost, in the same text, is one "that has both a fixed and variable component" — a fixed cost "does not change in total over the short term" while a variable cost "varies in direct proportion to the level of activity". The Saylor Academy managerial accounting text gives the identical formula and adds the caution that the method's "potentially significant weakness" is that "it only considers the high and low activity levels", so those points "may not represent the data set as a whole". Both texts present the high-low method as one of three estimation techniques alongside the scatter graph and least-squares regression, and both recommend the more sophisticated methods when the two extreme periods look unrepresentative. The estimate is valid only within the relevant range of activity actually observed.

How to use this calculator.

1. 01Pick one cost line and one activity measure, and keep both definitions identical across the two periods. The activity measure should be whatever actually drives the cost — units produced, machine hours, flight hours, orders shipped, support tickets — not simply revenue.
2. 02Find the period with the HIGHEST ACTIVITY, not the highest cost. This trips people up constantly: the method is defined on the extremes of activity, and choosing on cost instead will bias the slope. Enter that period's activity and its total cost.
3. 03Find the period with the lowest activity and enter the same two figures. Check both rows before continuing — if either was unusual for a reason unrelated to volume (a breakdown, a one-off insurance settlement, a strike, a stock write-off), the whole estimate inherits it. Substitute the next-nearest ordinary period instead.
4. 04Enter the activity level you want a cost prediction for. Keep it inside or close to the range between your two observations. The further outside you go, the more the estimate depends on an assumption of linearity that your data cannot support.
5. 05Read the cost equation. Variable cost per unit is the slope; fixed cost per period is the intercept; together they predict total cost at your forecast volume. Check the prediction against a period you did not use as an input — if it lands close, the linear model is holding.
6. 06Look hard at a negative fixed cost. It means cost grew faster than proportionally between your two points and a straight line is the wrong shape for this cost. Do not report it as a fixed cost; treat it as a prompt to look at more data points or to check whether a step change happened between the two periods.
7. 07Take the split onward. Variable cost per unit is the input to a contribution margin, and fixed cost per period is the input to a break-even calculation — both of which Quanta has dedicated calculators for.

The formula.

The calculator subtracts the low-activity cost from the high-activity cost and divides by the difference in activity, giving the variable cost per unit — the slope b. It then multiplies that rate by the high-activity level and subtracts the product from the high-activity cost, leaving the fixed component a. Because b is not rounded before this second step, the fixed cost derived from the high anchor is identical to the fixed cost derived from the low anchor, and the fitted line passes exactly through both observed points. Total cost at your forecast activity is a + b × x; cost per unit is that total divided by the forecast activity; and the two share percentages split the forecast total between its variable and fixed parts, always summing to 100%.

Rounding stage, and why it matters here more than anywhere else on the site: nothing is rounded part-way through. Textbook presentations typically round the variable rate to whole cents before deriving the fixed component, which is harmless when the division happens to be exact and wrong when it is not. Take a cost of $100 observed at 3 units and $0 at 0 units: the true rate is $33.333…, and a rate rounded to $33.33 would imply a fixed cost of $100 − $33.33 × 3 = $0.01 and a fitted line that misses the observed point by a cent. This module carries the rate at full decimal precision and rounds exactly once, at the point the result is returned, to ten decimal places. Both published worked examples on this page divide exactly, so neither could have exposed the difference — which is precisely why the awkward case is pinned by a test.

Using the worked example: high activity 5,900 units at $380,000, low activity 2,900 units at $200,000. The variable rate is ($380,000 − $200,000) ÷ (5,900 − 2,900) = $180,000 ÷ 3,000 = $60 per unit. The fixed component is $380,000 − ($60 × 5,900) = $380,000 − $354,000 = $26,000; anchoring on the low point instead gives $200,000 − ($60 × 2,900) = $200,000 − $174,000 = $26,000, the same figure. The cost equation is Y = $26,000 + $60x. At the forecast level of 6,000 units, predicted total cost is $26,000 + $360,000 = $386,000, cost per unit is $64.33, and the split is 93.26% variable against 6.74% fixed.

Four domain rules. The high-activity level must exceed the low-activity level, or there is no change in volume to divide by. The high-activity cost may equal the low-activity cost — that gives a purely fixed cost, variable rate zero — but it may not be lower, which would imply a negative variable cost and almost always means the two rows have been swapped. A forecast of zero activity returns the fixed cost with a cost per unit of zero rather than infinity. And a negative fixed component is returned, not rejected.

A worked example.

Frequently asked questions.

References& sources.

1. [1]OpenStax, Principles of Accounting, Volume 2: Managerial Accounting, §2.3 "Estimate a Variable and Fixed Cost Equation and Predict Future Costs" (OpenStax/Rice University, CC BY-NC-SA; retrieved 2026-07-29). Primary source for the high-low formula, the cost equation Y = a + bx, and the stated limitation that the method "is based on only two pieces of cost data". Contains the Regent Airlines fixture (23,000 flight hours/$90,000 high, 10,000/$64,520 low → $1.96 per hour, $44,920 fixed) reproduced in this page's tests. Open access.
2. [2]Heisinger, K. & Hoger, J., Managerial Accounting, §5.2 "Cost Estimation Methods" (Saylor Academy open textbook, CC BY-NC-SA; retrieved 2026-07-29). Independent second authority. States the same variable-rate formula, derives fixed cost by substituting the low point, and supplies the Bikes Unlimited fixture (5,900 units/$380,000 high, 2,900/$200,000 low → $60 per unit, $26,000 fixed, $386,000 at 6,000 units) used as this page's worked example and UI defaults. Also the source of the "may not represent the data set as a whole" caution. Open access.
3. [3]OpenStax, Principles of Accounting, Volume 2: Managerial Accounting, §2.2 "Identify and Apply Basic Cost Behavior Patterns" (retrieved 2026-07-29). Source for the fixed / variable / mixed cost definitions, the relevant-range limitation, and the statement that the same cost may be classified differently by different companies — which is why this page's intro states that classification is a judgement rather than a formula. Open access.
4. [4]Heisinger, K. & Hoger, J., Managerial Accounting, §5.1 "Cost Behavior Patterns" (Saylor Academy; retrieved 2026-07-29). Independent confirmation of the cost-behaviour definitions: a fixed cost "remains constant in total with changes in activity and varies on a per unit basis", a variable cost "varies in total with changes in volume of activity and remains constant on a per unit basis", and the relevant range is "the range of activity for which the cost behavior patterns are likely to be accurate". Open access.