Rounding Race — Play Free Online

Free mental maths drill — instant play, no signup, no download

Rounding Race shows a number and a place to round it to — nearest 10, 100, 1,000 or a decimal place — and asks for the result. It is a small skill that decides whether an estimate is useful, and it is the first step of almost every mental calculation worth doing: round, calculate, correct. Sixty seconds, one number at a time.

How to play

Pick whole numbers or decimals and press Start.

Each question names a number and the place to round to.

Type the rounded value; a correct answer advances immediately.

A miss shows which digit decided it.

Your score is numbers rounded correctly in the minute.

What it actually trains

Place value and the rounding rule, fast enough to use mid-calculation. Estimating is what catches a wrong answer before it is written down, and estimating well depends entirely on rounding to a sensible place without stopping to think.

Tips to score higher

Look only at the digit immediately to the right of the place you are rounding to. Everything after it is irrelevant.

Five and above rounds up; four and below stays. That is the whole rule.

Watch for the carry. Rounding 3,961 to the nearest hundred gives 4,000, not 3,900.

For estimating, round both numbers the same way and you will know which direction your estimate leans.

Frequently asked questions

How do you round to the nearest 10?

Look at the units digit. If it is 5 or more, round the tens up; if it is 4 or less, leave the tens as they are and make the units zero. 47 becomes 50, 43 becomes 40.

What happens when rounding forces a carry?

The carry moves left just like in addition. 3,961 to the nearest hundred is 4,000, because rounding the 9 up carries into the thousands.

Does 5 always round up?

In the convention used here and in most schools, yes. Some statistical work rounds half to even instead, but that is a specialist rule and this drill does not use it.

Why is rounding worth drilling?

Because estimating is the only cheap check you have. An estimate you can produce in a second will catch an answer that is out by a factor of ten, which is the error type that actually matters.