The Equivalating(???) odd numbers method or the fun system for counting votes in Sweden. (translation from Swedish wikipedia entry)
Only parties who achieve at least 4% nationally participate in the distribution of seats in parliament. The exception is for a party which gets 12% in a district which gets to participate in distribution of seats in that district but not anywhere else.

Comparison numbers for the parties are computed. The party with the largest comparison number gets the next seat to be allocated for an election district. The first comparison number for all parties is computed by dividing their respective number of votes by 1.4. After a party has been allocated a seat by having the largest comparison number, a new comparison number for that party is computed by dividing their number of votes by 2*n+1, where n is the number of seats already allocated that party. This process is repeated until all seats in the district have been allocated. The locally fixed seats (the seats allocated to each election district before the election takes place) are allocated to parties according to this model. Then there are some number (39) seats allocated nationally to achieve proportionality. This is also done according to the above method. (Can't find which comparison numbers they start with for this? Do they add up the final comparison numbers for all the districts for each party?) Once these "national proportionality distribution" seats have been allocated to the parties, they are assigned to election districts by using the final comparison numbers from the first local allocation. Except if the party didn't get any seats in a district. Then its comparison number is set to the number of votes gotten, not that number divided by 1.4.