• Total is the total number of people with that surname.
• Rank is the position in the list of names ordered by total (eg, a rank of 1 means that it's the most common name, and a rank of 10 means it's the tenth most common, etc).
• Frequency is the percentage of people with that surname.
• Per million people is the number of people with that surname per million of the population.

'N/A' indicates that we don't have data for this name (usually because it's quite uncommon and our stats don't go down that far). It doesn't mean that there's no-one with that name at all!

For less common surnames, the figures get progressively less reliable the fewer holders of that name there are. This data is aggregated from several public lists, and some stats are interpolated from known values. The margin of error is well over 100% at the rarest end of the table!

For less common surnames, the frequency and "per million" values may be 0 even though there are people with that name. That's because they represent less than one in a million of the population, which ends up as 0 after rounding.

It's possible for a surname to gain in rank and/or total while being less common per million people (or vice versa) as there are now more surnames in the USA as a result of immigration. In mathematical terms, the tail has got longer, with a far larger number of less common surnames.

Figures for top states show firstly the states where most people called COST live. This obviously tends to be biased towards the most populous states. The second set of figures show where people called COST represent the biggest proportion of the population. So, in this case, there are more people called COST in Ohio than any other state, but you are more likely to find a COST by picking someone at random in Alabama than anywhere else.