Children with periodic fever, aphthous stomatitis, pharyngitis, adenitis (PFAPA) -syndrome diagnosis will be randomised either to tonsillotomy (partial tonsillectomy) or 3 months follow up. At follow up visit 3 months after randomisation the children from either groups with ongoing symptoms will be sent to tonsillectomy.

Children with periodic fever, aphthous stomatitis, pharyngitis, adenitis (PFAPA) -syndrome diagnosis will be randomised either to tonsillotomy (partial tonsillectomy) or 3 months follow up.

In both groups the symptoms are monitored with symptom diaries. At follow up visit 3 months after randomisation the children from either groups with ongoing symptoms will be sent to tonsillectomy (total remove of palatine tonsil).

The design is sequential trial. From the previous trials we know that the effect of tonsillectomy is about 90% and that during follow up for six months as much as 50% of the patients will recover. Because the syndrome is rare and because the rescue operations of the tonsils (tonsillectomy in a person who has gone through tonsillotomy), the sample size has to be as low as possible. That is the reason why we chose sequential design, with assessments after every 8 patient.

Randomization is made in blocks of four. In sequential design we use the following assumptions:

• We are looking for a 40% absolute difference in cure rate between tonsillotomy (90%) and follow up (50%) groups.
• Two sided Whitehead design
• Maximum likelihood estimation (MLE)
• Five assessments (sample size 4+4, 8+8. 12+12, 16+16 and 19+19)
• type 1 error 5 % and power 80%
• The sample size will be from 8 to 38 patients depending on when the trial can be stopped along assessments.

Inclusion Criteria:

• Diagnosed PFAPA syndrome: regularly occurring fever episodes for 3-5 days for at least five times or for 6 months

Exclusion Criteria:

• prior tonsil surgery