Major　depressive　disorder　(MDD)　is　a　kind　of　mental　disease　which　has　characteristics　of　the　low　mood,　sense　of　worthless,　less　interest　in　the　surrounding　things,　sadness　or　hopeless,　slow　thinking,　intelligence,　language,　action,　etc.　The　aim　of　this　research　is　to　find　the　differences　between　entropy　values　and　ages,　genders　of　MDD　patients,　MDD　patients　and　healthy　controls.　Twenty-two　MDD　patients　(male　11;　age　18-65)　and　their　matched　healthy　controls　in　gender,　age,　and　education　are　examined　by　analyzing　(blood　oxygenation　level　dependent-functional　magnetic　resonance　imaging,　BOLD-fMRI)　signals　from　nonlinear　complexity　perspective.　As　the　BOLD-fMRI　signals　have　limited　time　resolution,　so　they　are　very　difficult　to　quantify　the　complexities　of　fMRI　signals.　We　extract　the　corresponding　signals　from　the　fMRI　signals.　The　complexities　of　the　age,　gender,　MDD　patients　and　healthy　controls　can　be　predicted　by　the　proposed　approach.　However,　information　redundancy　and　other　issues　may　exist　in　non-linear　dynamic　signals.　These　issues　will　cause　an　increase　in　computational　complexity　or　a　decrease　in　computational　accuracy.　To　solve　the　above　problems,　we　propose　a　method　of　fuzzy　approximate　entropy　(fApEn),　and　compare　it　with　sample　entropy　(SampEn).　The　addition　and　subtraction　under　different　emotional　stimuli　as　a　multi-task　are　used　to　coordinate　brain　sense　with　motion　control.　The　12-channel　fMRI　signals　are　obtained　involving　the　BOLD　signals　on　resting　signals　(about　24　s).　The　methods　of　the　fApEn　and　SampEn　are　proposed　to　deal　with　the　BOLD-fMRI　signals　in　the　different　ages　and　genders,　and　those　between　MDD　patients　and　healthy　controls　from　the　differences　between　fApEn　and　SampEn　of　different　genders,　main　effect　and　interaction　effect　analysis　of　fApEn　and　SampEn　measures,　regression　curve　between　entropy　and　age　of　the　whole　sample,　correlations　of　fApEn　and　SampEn　with　age,　fApEn-age　correlation　and　magnitude　in　gray　matter　and　white　matter,　multiple　regression　analysis　of　fApEn　with　age　for　the　whole　sample,　also　the　receiver　operating　characteristic　analyses　of　fApEn　and　SampEn,　the　relationship　between　fAPEn　and　N　aspects.　The　results　show　that　1)　the　complexities　of　the　resting　state　fMRI　signals　measured　are　consistent　with　those　from　the　Goldberger/Lipsitz　model:　the　more　the　health,　the　greater　the　complexity　is;　2)　the　mean　whole　brain　fApEn　demonstrates　significant　negative　correlation　(r　=　-0.512,　p　＜　0.001)　with　age,　SampEn　produces　a　non-significant　negative　correlation　(r　=　0.102,　p　=　0.412),　and　fApEn　also　demonstrates　a　significant　(p　＜　0.05)　negative　correlation　with　age-region　(frontal,　parietal,　limbic,　temporal　and　cerebellum　parietal　lobes),　there　is　non-significant　region　between　the　SampEn　maps　and　age;　3)　the　fuzzy　approximate　entropy　values　of　major　depressive　disorder　patients　are　lower　than　those　of　healthy　controls　during　resting.　These　results　support　the　Goldberger/Lipsitz　model,　and　the　results　also　show　that　the　fApEn　is　a　new　effective　method　to　analyze　the　complexity　of　BOLD-fMRI　signals.