You could do this over a month, as I do for my expenses.
Record your total mileage for 1 month. Total the amount in litres of fuel that you purchase during this period and divide by 3.79 to convert to gallons, (1 gallon = approx 3.79 litres) and divide the total mileage by the total fuel amount in gallons. E.g. If I did 786 miles in a month and put in 90 litres of fuel (23.75 gallons), then my miles/gallon would be 786/23.75 = 33 mpg
... or you could fill up the tank, write down the milage on the odometer, drive as you normally would untill you need petrol again, fill the tank up again and note how many gallons it takes and again note the odometer reading to see how many miles you've done since the last fill-up. Divide those miles by the number of gallons you've just tanked and you've got a pretty accurate mpg. You can repeat this procedure indefinitely and you'll also learn how different your mpg usage is between town use and open road traveling. Happy motoring.
Using wildwood's suggestion for just one filling can be fairly inaccurate. You need to do it over a period of time. Whenever I buy petrol, I record how much petrol I have used to fill the tank and record the mileage reading. That way I can, at any time, work out an average consumption over a period that I choose. Doing this, I have found that the consumption is consistently higher in the winter than it is in the summer.
Gef: I know! But more so because you have to heat the car, which takes energy, and use the heated rear screen more, which is far higher consumption that the lights etc.