Cost estimation of space science missions at APL over the past two dec
ades has been singularly successful in arriving at program costs that
are within a few percent of the actual costs at program completion. Th
e most recent example is the Near Earth Asteroid Rendezvous (NEAR) mis
sion which was estimated at approximately 112 million FY-92 dollars an
d came in at approximately three percent under the estimated cost. Thi
s demonstrated performance has been achieved without the benefit of a
formal cost model, such as those used in government and industry (GSFC
, MSFC, SAIC, etc.). In light of this performance, it is important to
understand the parameters that ate used in the cost estimating process
in an effort to quantify those elements in a program that are most im
portant to the final cost. We have identified a number of areas which
contribute to eventual cost performance; these include: (a) spacecraft
and mission complexity; (b) use of already-developed (facility-class)
instruments versus ''to be developed'' new instruments; (c) synergism
among programs being implemented concurrently; (d) program implementa
tion length; (e) design-to-cost practice for all major subsystems and
instruments without contingency; (f) lead engineer responsibility thro
ughout design, layout, fabrication, test, integration, and initial fli
ght operations; (g) designed-in quality and testability to minimize re
work; (h) incorporation of reliability and quality assurance engineeri
ng within the program structure; (i) minimization of documentation and
encouragement of oral and electronic communication as required. We ha
ve found that gross parametrization of costs such as the traditional w
eight, power, and length of the program commonly included in typical m
odels do not reliably predict actual costs. A methodology will be pres
ented, whereby, the elements identified above plus others are used to
describe the process implemented by APL in previous missions to genera
te cost estimates and to control costs. Actual data over several missi
ons during the past two decades will also be presented that illustrate
APL's cost performance while utilizing this methodology. Copyright (C
) 1997 Elsevier Science Ltd.